On Elementary Affective Decisions

Author’s Proof Carefully read the entire proof and mark all corrections in the appropriate place, using the Adobe Reader commenting tools (Adobe Help). Do not forget to reply to the queries. We do not accept corrections in the form of edited manuscripts. In order to ensure the timely publication of your article, please submit the corrections within 48 hours. If you have any questions, please contact [email protected].

Author Queries Form Query No.

Details required

Author’s Response

Q1

The citation and surnames of all of the authors have been highlighted. Please check all of the names carefully and indicate if any are incorrect. Please note that this may affect the indexing of your article in repositories such as PubMed.

✔

Q2

Confirm that the email address in your correspondence section is accurate.

✔

Q3

Please ask the following authors to register with Frontiers (at https:// www.frontiersin.org/Registration/Register.aspx) if they would like their names on the article abstract page and PDF to be linked to a Frontiers profile. Please ensure to provide us with the profile link(s) when submitting the proof corrections. Non-registered authors will have the default profile image displayed. Annette Kinder

Q4

If you decide to use previously published, copyrighted figures in your article, please keep in mind that it is your responsibility, as the author, to obtain the appropriate permissions and licenses and to follow any citation instructions requested by third-party rights holders. If obtaining the reproduction rights involves the payment of a fee, these charges are to be paid by the authors.

Q5

Ensure that all the figures, tables and captions are correct.

✔

Q6

Verify that all the equations and special characters are displayed correctly.

✔

Q7

Ensure, if it applies to your study, the ethics statement is included in the article.

✔

Q8

Ensure to add all grant numbers and funding information, as after publication this is no longer possible.

✔

Q9

Kindly confirm whether the Affiliations has been edited correctly.

✔

Q10

Please reduce “Keywords” to maximum of eight.

✔

Q11

Please include the following references in the reference list. Maddock et al., 2004; Kintsch, 2012; Ponz et al., 2014, 2016; O’Sullivan et al., 2015; Zeman et al., 2013.

✔

✔

✔

✔

Query No.

Details required

Author’s Response

Q12

Please confirm if the edit made in the sentence “Examples for most liked....(the torturer is spiteful)” is fine.

✔

Q13

We have changed the following references inside the text as per the reference list. Kindly confirm if this is fine. “Jose and Brewer, 1994” as “Jose and Brewer, 1984” “Leder, 2014” as “Leder and Nadal, 2014.”

✔

Q14

Please update the following references. Citron, 2016; Ullrich, 2016; Van den Hoven, 2016.

✔

Q15

We have changed “Citron et al., 2015” to “Citron et al., 2016” in the reference list as well as inside the text. Kindly confirm if this is fine.

Citron et al., 2015, 2016 is correct; I changed it

Q16

Kindly confirm whether the following references edited correctly. Hutchins, 1952; Jakobson, 1944.

✔

Q17

Please provide page range for “Jacobs, 2016a.”

Q18

Please provide volume number and page range for the following references. McQuire et al., 2016; Nicklas and Jacobs, 2016.

✔

Q19

Please cite “Van den Hoven et al., 2016” inside the text.

see line 224

164-174. doi 10.1075/ssol.6.1.08jac

HYPOTHESIS AND THEORY published: xx November 2016 doi: 10.3389/fpsyg.2016.01836 1

58

2

59

3

60

4

61

5

62

6

63

On Elementary Affective Decisions: To Like Or Not to Like, That Is the Question

7 8 9 10 11 12 13

Arthur M. Jacobs

15

18 19 20 21 22 23 24 25 26 27 28 29 30 31

35 36 37 38 39

Q2

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

*, Markus J. Hofmann and Annette Kinder 4

5

Department of Experimental and Neurocognitive Psychology, Freie Universität Berlin, Berlin, Germany, 2 Center for Cognitive Neuroscience, Freie Universität Berlin, Berlin, Germany, 3 Dahlem Institute for Neuroimaging of Emotion, Freie Universität Berlin, Berlin, Germany, 4 Department of General and Biological Psychology, Bergische Universität Wuppertal, Wuppertal, Germany, 5 Department of Education and Psychology, Freie Universität Berlin, Berlin, Germany

17

34

1,2,3

1

16

33

65 66 67 68 69 70

14

32

64

Edited by: Michael Burke, Utrecht University, Netherlands Reviewed by: Marcos Nadal, University of Vienna, Austria Matthew Pelowski, University of Copenhagen, Denmark

Perhaps the most ubiquitous and basic affective decision of daily life is deciding whether we like or dislike something/somebody, or, in terms of psychological emotion theories, whether the object/subject has positive or negative valence. Indeed, people constantly make such liking decisions within a glimpse and, importantly, often without expecting any obvious benefit or knowing the exact reasons for their judgment. In this paper, we review research on such elementary affective decisions (EADs) that entail no direct overt reward with a special focus on Neurocognitive Poetics and discuss methods and models for investigating the neuronal and cognitive-affective bases of EADs to verbal materials with differing degrees of complexity. In line with evolutionary and appraisal theories of (aesthetic) emotions and data from recent neurocognitive studies, the results of a decision tree modeling approach simulating EADs to single words suggest that a main driving force behind EADs is the extent to which such high-dimensional stimuli are associated with the “basic” emotions joy/happiness and disgust. Keywords: neurocognitive poetics, elementary affective decisions, valence, liking, beauty, emotion, neuroaesthetics, ludic reading, decision tree modeling, basic affective tone

Received: 22 September 2016 Accepted: 07 November 2016 Published: xx November 2016 Citation: Jacobs AM, Hofmann MJ and Kinder A (2016) On Elementary Affective Decisions: To Like Or Not to Like, That Is the Question. Front. Psychol. 7:1836. doi: 10.3389/fpsyg.2016.01836

72 73

Q1 Q3 Q9

74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94

Q10

95 96

*Correspondence: Arthur M. Jacobs [email protected] Specialty section: This article was submitted to Theoretical and Philosophical Psychology, a section of the journal Frontiers in Psychology

71

“How then did it work out, all this? How did one judge people, think of them? How did one add up this and that and conclude that it is liking one felt, or disliking?” – Virginia Woolf, To the Lighthouse

97 98 99 100 101

INTRODUCTION

102 103

Deciding whether we like or dislike something/somebody, i.e., – in terms of psychological emotion theories – whether it has positive or negative valence is perhaps the most ubiquitous and basic affective decision of daily life. We constantly make such liking decisions within a glimpse – not only in Facebook –, and often without expecting any obvious benefit or knowing the exact reasons for our judgment. Here we would like to introduce the term elementary affective decisions (EADs) for this kind of ubiquitous decisions that entail no direct overt reward. While EADs are made with regard to all types of perceived objects in daily life (Lebrecht et al., 2012), here we focus on EADs concerning verbal materials in the context of literature reception and reading from a Neurocognitive Poetics perspective, i.e., the investigation of the neuronal, experiential, and behavioral effects produced by literary texts (Jacobs, 2015b). This has several advantages. First, there exist well validated databases allowing to quantify not only the valence of written words

Frontiers in Psychology | www.frontiersin.org

1

November 2016 | Volume 7 | Article 1836

104 Q6 105 Q7 106 107 108 109 110 111 112 113 114

Jacobs et al.

115 116 117 118

Q11

119

2003 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155

On Elementary Affective Decisions

and texts, but also various other features that may interact with valence in the process of EADs, e.g., stimulus arousal or imageability. Second, the majority of empirical studies on EADs has used words as stimuli thus providing a solid basis for review and model development (e.g., Maddock et al., 2004; Võ et al., 2006; Unkelbach et al., 2008; Jacobs et al., 2015; Sylvester et al., 2016). Finally, the focus on words having quantifiable features allows computational modeling of EADs testing the main hypotheses proposed in this paper. The central working hypothesis we pursue in this paper is that the computed valence of single words and the mostly preconscious and preverbal EADs accompanying word recognition forms the basis of pleasurable aesthetic literary experiences with more complex verbal materials. A more specific hypothesis states that valence is a semantic superfeature that results from a yet unknown integration of both experiential and distributional data, at least partially represented in associative activation patterns of semantic networks, as assumed by the semantics theory of Andrews et al. (2009). A valence value would thus be computed from (1) neural activation patterns distributed over the sensory-motor representations of a word’s referents (experiential aspect) and (2) the linguistic company the words keep (Harris, 1951), i.e., the size and density of their context (distributional aspect; Jacobs et al., 2015), as computationally modeled using co-occurrence statistics (Hofmann and Jacobs, 2014). In favor of this view, Westbury et al. (2014) recently showed that valence ratings of words can be predicted by their associations to a selected set of emotion labels, derived from theories of basic emotions (cf. also Hofmann and Jacobs, 2014). More recent behavioral and neurocognitive evidence stemming from experiments using words of neutral valence and bivalent noun-noun compounds (i.e., novel words created by fusing two nouns that have different valences, such as BOMBSEX) supports this view (Kuhlmann et al., 2016). In the following we start with a short review of theories of emotion, aesthetics, and reading relevant for our purposes. A section on “Methods and Materials” used in the study of EADs then prepares the next section which introduces factors determining the liking and neuronal correlates of verbal materials with increasing degrees of complexity. A final section presents formal models of EADs with different degrees of complexity.

is a life-long process starting from the earliest days, e.g., through 172 discovering new meanings offering the promise of novel insights, 173 enjoying word plays, sounds and rhythms, self-rewarding feelings 174 of suspense, surprise and vicarious joy, fear or disgust. As argued 175 elsewhere, later joyful or enlightening encounters with poetry are 176 likely to be rooted in early life experiences with micropoetry. 177 These include preverbal and preliterate (micro)poetic episodes, 178 e.g., lullabies, nursery rhymes, and word games, making children 179 enjoy rhyme and delight in rhythm and repetition (Freud, 1907; 180 Jakobson, 1944; Miall and Dissanayake, 2003; Wainwright, 2011; 181 Jacobs, 2015c; Jacobs and Kinder, 2015). 182 It also has been argued that from early, preliterate childhood 183 on, word games play an important role in all kinds of 184 incidental or ludic learning activities, as well as in emotion 185 regulation and mood management and that these activities are 186 the precursors of ludic reading and aesthetic liking accompanying 187 many subsequent literary experiences (Jakobson, 1944; Jacobs, 188 2015c; Jacobs et al., 2015; Pompe, 2015). Burke (2013, p. 208) 189 wonderfully expresses this: “when he/she subconsciously longs 190 for the style figural, rhythmic rewards. . ..that sense of balance, of 191 perfection, of exactitude (in the context of adults reading Joyce’s 192 short story “The dead”), or “how strong the subconscious desire 193 is to be moved by emotive literary style fragments.” But this raises 194 the question of how children and young adults acquire these 195 literary longings and desires. With regard to children reading 196 Dr. Seuss’ Horton Hears a Who, Boyd (2001, p. 204) puts it this 197 way: “He (i.e., the author) can count on children’s pleasure that 198 someone has made up a story for them to respond to, as a parcel 199 of pleasure, a gift of attention.” 200 Within the field of neurocognitive poetics (Jacobs, 2015a, 201 2016b; see also Burke, 2015) we are only beginning to understand 202 the potential causes and effects of ludic reading. One line of 203 theorizing trying to explain what motivates readers to pass time 204 with written stories or poems highlights the role of immersion. 205 The multidimensional construct of immersion, as first proposed 206 by the film theoretician Bela Balázs, and then applied to literature 207 reception by Schrott and Jacobs (2011), is perhaps best described 208 as getting lost in a book (Nell, 1988). Other metaphors than Balázs’ 209 are popular in the literature, e.g., absorption or transportation, 210 but at present both the conceptual work and empirical data base 211 is not enough developed to allow sharp distinctions (Jacobs and 212 Schrott, 2015; Jacobs and Lüdtke, 2017). What seems clear is 213 that young and adult readers long for immersion into novels 214 like Harry Potter and a few pioneering studies have begun to 215 investigate the neurocognitive and –affective underpinnings of 216 this phenomenon (Hsu et al., 2014; Lüdtke et al., 2014). It is 217 less clear whether readers’ longing for poetry like Shakespeare’s 218 or Pushkin’s, as well as highly foregrounded narratives from 219 Joyce or Proust is based on the same kind of neuropsychological 220 processes or linked more to processes of aesthetic appreciation, as 221 proposed by the Neurocognitive Poetics Model (NCPM; Jacobs, 222 2011, 2015a, 2016b; cf. also Miall, 1989; Lüdtke et al., 2014; 223 et al., 2016 Nicklas and Jacobs, 2016; Willems and Jacobs, 2016). cf. also Van den Hoven224 Although investigating the liking of complex and extensive 225 verbal materials like novels is a worthwhile scientific activity 226 (Burke, 2011), empirical studies on the possible foundations of 227 pleasurable literary experiences, e.g., studying EADs with single 228

156 157 158

LIKING WORDS AND LUDIC READING

159 160 161 162 163 164 165 166 167 168 169 170 171

Reading is indeed one of those ubiquitous activities often chosen for pleasure, and the mental and neuronal processes underlying ludic reading and literary experience in general have been the object of increasing research efforts (e.g., Nell, 1988; Anz, 1998; Schrott and Jacobs, 2011; Jacobs, 2015c; Jacobs and Schrott, 2015). Although little is known about the processes underlying children’s learning to like things, in particular regarding pleasurable literary experiences, there are impressive testimonies by children explaining, for example, why they find even single words beautiful (Limbach, 2004; Jacobs et al., 2015): A phenomenon that could be termed episodes of micropoetry (cf., Jacobs, 2015c; Jacobs and Kinder, 2015). Learning to like language


2


Jacobs et al.

229 230 231 232


words or sentences, remain a valuable source of insight, as do theoretical analyses in the light of more general theories on emotion or aesthetics. In the following we briefly discuss some relevant literature.

elicited nor enhanced by instructions to report the outcome of these judgements. In sum, human neuroimaging studies suggest that brain areas associated with aesthetic responses to artworks overlap with those that are linked to the appraisal of objects of evolutionary importance (such as desire for food or attractiveness of faces). A plausible working hypothesis thus is to assume that artworks have co-opted the neural systems that subserve these kinds of adaptive assessments rather than having evolved a distinct type of neural processing (Brown et al., 2011). In the emerging field of neurocognitive poetics, not addressed in Brown et al.’s meta-analysis, the Panksepp-Jakobson Hypothesis (Jacobs, 2015b) expresses this similarly in an attempt to bridge the language-emotion gap. This concerns the gap between neurobiological theories of emotion, as perhaps best represented by Panksepp’s (1998) core affect systems theory, and complex (psycho-)linguistic models, as exemplified by Jakobson’s (1960) extended version of Bühler’s (1934) ‘organon model’ of language functions (Jacobs et al., 2015; see Koelsch et al., 2015, for a theory linking emotion and language). Stated simply, the PankseppJakobson Hypothesis submits that since evolution had no time to invent a proper affective system for art reception, even less so for reading, the emotional and aesthetic processes we experience when reading must be somehow linked to the ancient neuronal affect circuits we share with all mammals (Panksepp, 1998). There is accumulating evidence for this hypothesis stemming from work on neurocognitive poetics generalizing across various stimulus materials, subjects, and tasks (for review, see Jacobs, 2015b). In sum, it seems safe to say that liking, i.e., experiencing pleasure in terms of emotion theories – be it with positive or negative valence, or a mixture of both –, plays a role in feelings of beauty, most certainly when the latter are compressed into numerical ratings. To say it in the words of Silvia (2009, p. 48): “The psychology of aesthetic experience is eerily close to the psychology of how much novices say they like something.” However, the heterogeneity of notions, e.g., aesthetic experience, aesthetic episode, aesthetic appraisal, aesthetic judgment, aesthetic feeling/emotion, aesthetic evaluation etc., methods and materials used to assess felt beauty make it difficult to draw more precise conclusions at this stage (cf. Güçlütürk et al., 2016; Marin et al., 2016; Pelowski et al., 2016). In the following sections, we examine methods used to measure EADs in the context of verbal materials of differing degrees of complexity.

233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262

delete

263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285

THE ROLE OF VALENCE AND BEAUTY IN THEORIES OF EMOTION, AESTHETICS, AND READING Historically, psychology has investigated EADs or liking judgments from two major perspectives: emotion theories (e.g., Wundt, 1896) and theories of aesthetic judgment or beauty evaluation (e.g., Eysenck, 1941; Berlyne, 1971; Leder et al., 2004; Jacobsen, 2006; Nadal, 2013; Chatterjee and Vartanian, 2014; Leder and Nadal, 2014). Both perspectives produced a wealth of models and methods utilizing countless stimuli and tasks, but have been pursued relatively independently of each other without much theoretical and methodological cross-over (but see Pelowski et al., 2016). As a consequence, the issue whether liking judgments tap into the same underlying processes when participants are, for example, asked to rate the valence of emotion inducing pictures or words (i.e., in emotion-related research) as compared to when they are invited to rate the beauty of paintings, music, proverbs, or poems in research on (neuro-)aesthetics, still is an open one (cf. Marin, 2015; Omigie, 2015). Finding something/someone pleasurable and finding something/someone beautiful are logically independent, e.g., when liking something/someone because of other aspects than aesthetic ones, or attributing aesthetic value to a piece of art without finding it beautiful (i.e., without producing an experience of aesthetic pleasure). However, psychologically pleasure and beauty seem intertwined, e.g., in appraisal theories (e.g., Silvia, 2009) or semantic differentials (Berlyne, 1971) that they are often indistinguishable (i.e., hedonic terms). Empirically one also often finds a high correlation between liking and beauty judgements (e.g., Lüdtke et al., 2014). This fits with the classical notion shared by scholars as different as Kant, Gadamer, or Ramachandran that pleasure (associated with valence) is a necessary component of aesthetic feelings (Brown et al., 2011; Jacobs, 2015b). A central goal of neuroaesthetics thus consists in naturalizing aesthetics by grounding aesthetic feelings in general theories of emotion (Brown et al., 2011). The latter authors argue that aesthetic processing is, at its core, the appraisal of the valence of perceived objects1 . This (or a similar type of) appraisal is not limited to artworks (including verbal ones) at all, but is applicable to all types of perceived objects in daily life (Lebrecht et al., 2012). A recent meta-analysis on the topic (Kühn and Gallinat, 2012) concludes that subjective EADs are directly related to brain regions that have been described as part of the reward circuitry (e.g., medial orbitofrontal cortex, ventral striatum), and that the evaluation of likability is an automatic process that it is neither

287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331

METHODS AND MATERIALS FOR MEASURING EADS

1

Note that aesthetic processing is best thought of as a binary phenomenon, with both positive and negative counterparts. Thus, negative-valenced emotions such as dislike and disgust are just as much aesthetic emotions as are awe and ecstasy (Brown et al., 2011).


286

333 334 335

The simplest task used to measure EADs is the valence decision task (VDT2 ) asking participants to explicitly decide whether a stimulus is positive or negative (pleasant/unpleasant; beautiful/ugly), either while recording their response times (RTs)

336

2

341

Synonyms used in the literature are valence identification or classification, valence judgement, affective decision, or aesthetic judgement task.

3

332


337 338 339 340 342

Jacobs et al.

343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399


structure and affect EADs to words: three affective-semantic variables (valence, arousal, imageability), three (sub-)lexical ones (word frequency, number of syllables, neighborhood density), five discrete emotion variables (joy/happiness, fear, anger, sadness, disgust; Briesemeister et al., 2011, 2014, 2015), and three embodiment ones (taste, grasp, move). Among the discrete emotions, joy and disgust may play a special role for liking decisions since from a Darwinian perspective they form a logical pair, two sides of a medal, related to incorporation of attractive food (or mating partners) on the one hand (joy) and rejection of aversive food on the other (disgust). These sociobiological functions of incorporation and rejection can be generalized to symbolic stimuli thus forming a potential evolutionary, discrete emotion-related basis of liking and disliking (see section on modeling at the end of this paper). Similarly, joy and disgust also play a role in appraisal theories of aesthetic emotions (e.g., Silvia, 2009; cf. also Pelowski et al., 2016). Altogether, the above considerations suggest that To Like Or Not To Like a single word is already a complicated multidimensional affair relativizing the above discussion about the dimensionality of the valence construct. So how can we explain EADs to multiword artifacts and verbal artworks such as metaphors, proverbs, idioms, sentences, passages, or entire stories and poems (Jacobs, 2016b)?

or ratings, or both (e.g., Hill and Kemp-Wheeler, 1989; Maddock et al., 2003; Kawabata and Zeki, 2004; Võ et al., 2006, Võ et al., 2009; Unkelbach et al., 2008; Jacobs et al., 2015; McQuire et al., 2016; Sylvester et al., 2016). In the more complex valence rating task (VRT) participants typically are asked to rate the valence of stimuli on a scale (from 3 to 3, or 1 X, where X can be 5, 7, or even greater numbers). Both the VDT and VRT have produced ubiquitous phenomena with various types of stimuli (e.g., pictures, words, faces, paintings): (1) the inversely, slightly asymmetric U-shaped function relating RTs in the VDT to stimulus valence (e.g., Võ et al., 2006) or beauty (Kawabata and Zeki, 2004), and (2) the asymmetric U-shaped function relating valence to arousal ratings in the VRT (e.g., Võ et al., 2009; Schmidtke et al., 2014). We have discussed theoretical reasons for these phenomena elsewhere (Sylvester et al., 2016; cf. Ashby and Isen, 1999; Koch et al., 2016) and will not go into details here. For the present purposes, however, two points are important: first, negative stimuli are usually more arousing than positive ones, and, second, positive ones typically are processed faster. One lesson to learn for studies of EADs then is that the factor arousal must not be neglected or ignored, as is often the case in studies of reading. The second, more tricky issue concerns the dimensionality of the construct valence. Wundt’s (1896) original conception of valence as a bipolar dimension had an important methodological (or measurementtheoretic) implication: valence typically is rated on bipolar scales (i.e., from very negative to very positive or very pleasant to very unpleasant). Thus, when a word or poem is judged as more pleasant, it automatically becomes less unpleasant. However, this notion has been challenged by theories that conceive valence as a bivariate construct which would require a different way of measuring it (e.g., Norris et al., 2010; Briesemeister et al., 2012). The issue is crucial because if a stimulus such as a word, face, or object can have positive and negative valence at the same time, i.e., is affectively ambivalent, this makes a huge difference especially with respect to aesthetic theory which has often highlighted the role of mixed emotions for aesthetic liking. For verbal materials, there is some preliminary evidence in favor of the bivariate perspective on valence coming from a study on words taken from the “Berlin Affective Word List” (BAWL; Võ et al., 2006, Võ et al., 2009). However, Briesemeister et al.’s (2012) study seems to be the only one addressing this issue with word materials, and thus the literature remains inconclusive resulting in several competing (descriptive) models of the VDT (Jacobs et al., 2015). The second big issue concerns factors determining the valence of a stimulus and driving an EAD. What exactly is it that, for instance, makes people judge a word as pleasant or unpleasant, beautiful or ugly? Words are quasi-ideal stimuli to tackle this issue, because more than 50 quantifiable word features are known (Graf et al., 2005) that can be examined with regard to their role in valence judgments (Jacobs et al., 2015). As outlined in the “Introduction,” the latter authors recently proposed word valence to be a semantic superfeature resulting from a combination of experiential and distributional aspects. They identified 14 variables that contribute to this mixed semantic


400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

FACTORS INFLUENCING AND NEURONAL CORRELATES OF EADS TO VERBAL MATERIALS The next section reviews features of words and text segments that have been shown to affect EADs to verbal materials. These features per se obviously provide neither a necessary nor a sufficient criterion for a piece of text to be judged as beautiful or poetic, given the intricate dynamics of the text-reader-context nexus (Jacobs, 2015c, 2016a) and the hierarchical and dynamic nature of beauty judgments (Kintsch, 2012).

SINGLE AND COMPOUND WORDS

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440

When participants quickly (i.e., automatically) decide whether they like a word or not, as in the VDT, a complex of more than 10 variables accounts for variance in RTs (see above). To give an illustrative example from research with children (8–12 years old), the five most liked words in the kidBAWL database (Sylvester et al., 2016) were: NATURE, MAMA, GIFT, SMILE, and FRIEND; those least liked were: VIOLENCE, MURDER, CADAVER, DECEPTION, and STEAL. As concerns the example MAMA, the phylo- and ontogenetically determined optimal combination between euphony and eusemy can explain its automatically perceived beauty (Jacobs, 2015c), but when participants rate the beauty/ugliness of words in a less automatic, more time-consuming procedure, it is likely that even more (and other) factors enter into the mental equation. Regrettably, empirical work on the beauty of words is almost absent, but some preliminary data from a study on 450 words chosen from

4


441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456

Jacobs et al.

457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511


databases like “The most beautiful German Word” (Limbach, 2004), dictionaries of German adolescent language, and the BAWL, suggest that reference to phenomena from nature (e.g., animals, flowers, rainbow etc.) and states/objects of wellness (e.g., coziness), together with high (positive) valence, familiarity, and imageability values, and low arousal values, make words appear more beautiful3 . The most beautiful words in the sample were LIBELLE (dragonfly), MORGENRÖTE (aurora), and MITTSOMMERNACHT (midsummernight). In contrast, the ugliest words in that sample were almost all swear words associated with genitalia (Jacobs et al., 2015: supplementary materials). As more fully discussed in Jacobs et al. (2015), the 9-year old Sylwan Wiese explains why the word LIBELLE (dragonfly) is the most beautiful for him by refering to three cues: a perceptual one (the wobbling, a motion which he loves watching), a phonological one (the Ls which make the word glide so well on his tongue), and an affective-semantic feature (no fear, because the word itself expresses the feeling evoked by the wobbling ensuring that one is not afraid of these insects). Thus, for this child and perhaps also for others it seems that both associations with discrete emotions and embodied cognitions play a role in aesthetic appreciations of words. In terms of Kintsch’s (2012) computational model of beauty, words like MAMA or LIBELLE may be likeable and beautiful because the divergence between the distinctive codes for the perceptual-imageable, (sublexical) phonological and affective-semantic parts and the (lexico-semantic) code for the word as a whole is small, thus giving it a good Gestalt. In one of the first neuroimaging studies examining word valence effects in an implicit task not requiring any attention to affective or aesthetic word properties (lexical decision), Kuchinke et al. (2005) found selective activations for positive words in brain regions associated with reward and semantic retrieval (orbitofrontal cortex; middle temporal gyrus). This is in line with the above mentioned neuroaesthetic studies and, moreover, with the idea that liking words is related to a more fluent retrieval from semantic memory (i.e., their familiarity). If EADs are (also) a function of the perceiver’s processing dynamics (Reber et al., 2004) – i.e., the more fluently readers can process a word or text segment, the more positive is their aesthetic liking response – then the more elaborated and interconnected (cohesive) schemata of positive words or objects (Ashby and Isen, 1999; Sylvester et al., 2016) may be a key factor explaining liking. Recent neurocognitive and -computational evidence supports this account by showing that pleasant words take advantage of a primarily left amygdala-mediated enhanced perceptual processing (Herbert et al., 2009) and that they provide more and denser semantic long-term associations than neutral or negative words (Hofmann and Jacobs, 2014). The link between memory associations and liking (positive valence) can be further explained in terms of complementary learning systems theory (Kumaran and McClelland, 2012), and by the hypothesis that the inferential function of the hippocampus can create a lucid positive feeling that may be figuratively described by the ‘light of understanding’ (Hofmann and Kuchinke, 2015).

Thus, the neurocognitive function of recurrent feedback from the hippocampus to the temporal cortex can be circumscribed by Oscar Wilde’s word “anything is good that stimulates thought” (Hyde, 1962, p. 108). There are also studies providing data on the liking of more complex words, e.g., noun-noun compounds (NNCs) which are a very productive word class in German: One can create countless neologisms by coupling noun pairs (or, to a lesser extent, adjective/verb-noun pairs) varying along several theoretical dimensions such as familiarity, literality-metaphoricity (Forgács et al., 2012), or valence (Jacobs et al., 2015; Kuhlmann et al., 2016). With regard to the issue at hand, NNCs are indeed very interesting, because they allow to couple nouns of opposite valence, thus creating bivalent words that challenge EADs by creating a decision conflict. Jacobs et al. (2015) presented preliminary evidence that such newly created bivalent NNCs (e.g., bombsex) prolonged RTs in a VDT as compared to univalent NNCs (e.g., pimple-horror), likely due to the decision conflict interfering with meaning construction. The most liked NNCs in Jacobs et al.’s corpus were the neologisms: “Glücksgenuss” (happiness-enjoyment), “Fabelliebe” (fablelove), and “Traumfreiheit” (dream-freedom), the least liked were: “Gewaltgeschwür” (violence-abscess), “Foltervorwurf ” (torture-reproach), and “Schurkentyrann” (rascal-tyrant). Neurocognitive experiments examining the processing of NNCs like “DUFTGESANG” (fragrance-chant; Kuhlmann et al., 2016) suggest that they require additional semantic integration work, as correlated with increased left inferior frontal gyrus activity as a function of both their familiarity and figurativeness (Forgács et al., 2012). To what extent the figurativeness or metaphoricity of NNCs or other words contributes to their liking has not yet been investigated, but is an interesting issue for future studies (cf. section on idioms below). More generally concerning neural correlates of word valence, positivity is neuroanatomically most often associated with the basal ganglia including the ventral striatum, left frontal pole, medial orbitofrontal cortex, ventromedial prefrontal cortex, posterior cingulate cortex, and supplementary motor area, whereas negativity is rather associated with insula, right amygdala, periaqueductal gray, right dorsal anterior cingulate cortex, left orbitofrontal cortex, dorsomedial prefrontal cortex, and deep cerebellar areas. Amygdala and anterior insula activations – or an amygdalar-hippocampal network – are specifically associated with word arousal (see Citron, 2012, and Jacobs et al., 2015, for reviews). The anterior insula seems to play a key role in disliking words. Ponz et al. (2014) used ugly or disgusting words taken from the BAWL to test whether brain regions involved in processing emotional information in general (e.g., in faces, pictures, or smells) are also in charge of the processing of emotional information in words. Comparing the processing of words like VIRUS, AMPUTATION, or PISS to that of control stimuli, they found support for this specific variant of the Panksepp-Jakobson Hypothesis. In sum, simplifying an ever increasing heterogeneous complex of neurocognitive results, one could propose that EADs to words are primarily associated with orbitofrontal networks including the dorsolateral prefrontal cortex (Herrington et al., 2005) and the insular cortex.

512 513

3

Valence was found to be the strongest predictor for beauty ratings (R2 lin = 0.77).


5


514 515

516 adage 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570

2013

Jacobs et al.

571


MULTIWORD EXPRESSIONS

language comprehension and conceptual representations, i.e., the left pre- and post-central gyri and the right superior temporal gyrus, whose activity was enhanced by both positively and negatively valenced stimuli, but not in the typical “emotionrelated” areas discussed above.

572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591

2015 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612

2015 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627

Multiword expressions that form idioms or proverbs (see Cacciari, 2014, for review) have been examined with regard to their likeability and beauty in both behavioral and neurocognitive studies (Bohrn et al., 2012a,b, 2013; Citron and Goldberg, 2014; Citron et al., 2016a,b). 2015, 2016


6

630 631 632 634

According to some authors, the little pearls of wisdom called proverbs have a special integrative potential reaching into the structure of human awareness where cognition, emotion, and volition often lose touch with each other to the detriment of all three (Hernadi and Steen, 1999). Their poetic and often anarchic offspring, so-called anti-proverbs, represents a special case of linguistic adaptation – more or less artful alternations of original proverbs like A Rolling Stone Gathers Momentum (Mieder, 2004; Jacobs, 2015c; Nicklas and Jacobs, 2016). In Bohrn et al. ’s (2012b, 2013) neurocognitive studies comparing familiar vs. unfamiliar proverbs with anti-proverbs, the most beautiful proverbs were familiar ones like “Wissen ist Macht” (knowledge is power) or “Wer wagt, gewinnt” (who dares, wins) thus outperforming the witty, artfully twisted antiproverbs through the powers of familarity and processing fluency (Reber et al., 2004). In terms of Kintsch’s (2012) model of beauty, proverbs might represent a verbal construction that minimizes complexity by averaging and compressing a (deeper) meaning over many instances of utterances/sentences that may not be quite prägnant or beautiful in themselves, thus ironing out the imperfections of particular instances and yielding an idealized verbal image. Anti-proverbs like “Mens sana in campari soda” (Nicklas and Jacobs, 2016) are interesting stimuli because they present a nice example of how background and foreground features can be combined in a single sentence (Jacobs, 2015b): Due to their multiple rhetoric features (phonological similarities like rhyme/alliteration, repetition and parallelism, meter, brevitas, i.e., artful shortness/prägnanz, or ellipses), all proverbs can be considered foreground elements of language if seen against a background of literal, i.e., non-rhetorical, nonfigurative control sentences. However, while the memory of the original proverb (e.g., ‘All roads lead to Rome’) provides familiar background information, the one-word change in antiproverbs like ‘All sins lead to Rome,’ creates a foregrounding effect and a subjectively felt tension, perhaps similar to puns. It also produces affective and aesthetic responses, the neuronal traces of which can be measured using fMRI. Thus, Bohrn et al. (2012b) observed data compatible with the interpretation that anti-proverbs evoking two contrasting responses that have to be related (that of the familiar proverb and the novel word), required a greater semantic integration effort, as mirrored by stronger ventral inferior frontal gyrus activation than for control stimuli (much as with the aforementioned bivalent NNCs; Forgács et al., 2012; Kuhlmann et al., 2016). Moreover, concurrent increased medial orbitofrontal cortex and striatal activation likely reflected the rewarding aspect of successful semantic integration (a kind of “Aha” experience; Bar et al., 2006; Hofmann et al., 2014) and supports both the findings and views of the naturalizing aesthetics studies discussed above.

Idioms are word strings whose global meaning cannot generally be inferred merely on the basis of the meaning of the constituent words (i.e., semantic transparency), and therefore has to be retrieved from semantic memory (Cacciari, 2014). Idioms like “Pfeffer im Hintern haben” (to have pepper in the ass) thus differ from proverbs, which, as literally and figuratively true statements, usually are temporarily undefined full sentences, signaled by specific grammatical, phonetic, and/or rhetorical patterns, or by a binary structure (theme/comment). Although some idioms can diachronically come from metaphors, they also differ from them, since metaphors (even the most frozen/dead ones) do not possess a unique standardized meaning and can convey more than one meaning depending on context (Citron et al., 2016b). The most liked idioms in the German database of over 600 stimuli by Citron et al. were: “vor Freude strahlen” (to beam of joy), “im siebten Himmel sein” (to be in seventh heaven), and “auf Wolke sieben schweben” (to float on cloud seven). Besides reproducing the above-mentioned asymmetric U-shaped function relating valence to arousal ratings observed with words (suggesting that idioms are affectively semantically processed like single words), the data of Citron et al. also allow to examine to what extent (rated) figurativeness contributes to the liking (rated valence) of multiword expressions. To illustrate contrasting degrees of figurativeness of German idioms consider the following examples for low figurativeness: “Das ist keine Kunst” (literal translation: this is no art; meaning: this is not difficult), or “keinen Pfennig mehr haben” (literal translation: to have no more cent; meaning: to be broke/out of money). In contrast, idioms exposing high figurativeness ratings are: “einen Kater haben” (literal translation: to have a tomcat; meaning: to be hung-over), or “grün hinter den Ohren sein” (literal translation: to be green behind the ears; meaning: to be immature/unexperienced). The results in Table 8 of Citron et al. (2016b) suggest that EADs to idioms depend on their familiarity, arousal values, and figurativeness. Since figurativeness also correlated negatively with both concreteness and semantic transparency ratings (and positively with idiom length, i.e., log number of words/idiom), the data of this study tentatively suggest that idioms are liked more the better they are known and understood (i.e., familiarity), very likely reflecting the ubiquitous relation between processing fluency and aesthetic pleasure discussed above. Citron et al. (2016a, in revision) recently compared idiomatic and literal sentences of negative, neutral, and positive valence in an fMRI study. They found that idioms elicited significantly enhanced activation of the left amygdala, and bilateral inferior frontal gyrus, the right temporal cortex, and the right precentral gyrus. Valence effects were seen in brain areas associated with

629

633

Proverbs

Idioms

628


635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684

Jacobs et al.


In sum, various stimulus features and neuronal networks are involved in EADs to multiword expressions. At the experiential and behavioral levels of observation, the superfeature 687 valence together with arousal, familiarity/fluency, and minimal 688 complexity appear to be key factors. At the neuronal level, apart 689 from the reward circuitry networks associated with emotional 690 engagement, semantic integration. and language processing in 691 NSERT: , general play a role (Bohrn et al., 2012a). 692

is romantic); least liked were sentences such as “the waste is toxic”, or “Der Folterer ist gehässig” (the torturer is spiteful). The behavioral results obtained in an implicit task (sentence verification), that in contrast to Bestgen’s study did not require any attention to word valence and was context-free (i.e., no tale or story containing them), as well as in a second explicit rating task did not support the null-model. As concerns the latter task, the authors observed an interactive effect falsifying the linear combination model: sentences with positive and neutral adjectives after negative nouns (e.g., “Der Einbrecher ist schlau”/the burglar is smart) were evaluated more negatively than sentences with positive and neutral adjectives after positive nouns (e.g., “Die Geliebte ist sinnlich”/the lover is sensual). This negativity bias for negative adjectives suggests an interaction of the general human negativity bias reported in the emotion processing literature discussed above and the syntactic role of words during sentence processing4 . The authors interpreted this bias in terms of the above discussed asymmetry in semantic representations of negative vs. positive words: negative words are less homogenous, i.e., have less elaborated and cohesive representations (Ashby and Isen, 1999; Koch et al., 2016; Sylvester et al., 2016). Lüdtke and Jacobs (2015) therefore assumed that semantic integration and coherent situation model building for sentences with two negative words could be harder compared to sentences with two positive words. Since the condition in which the shortest verification times were observed, i.e., congruent positive nouns and adjectives, also received the highest valence ratings, Lüdtke and Jacobs also hypothesized that processing fluency and meaning construction are positively related to automatic valence evaluation, i.e., EADs. This supports the relation between perceptual and conceptual processing fluency and aesthetic pleasure, the link between memory associations and liking already discussed for single words, compounds, idioms, and proverbs. For sentences with two positive words, semantic activation can spread across better elaborated and connected associative pathways, and thereby elicit a positivity bias or positivity superiority effect during meaning construction. In line with this, Havas et al. (2010) had already observed that reading times for what they called “happy” sentences like “You spring up the stairs to your lover’s apartment,” or “Finally, you reach the summit of the tall mountain” were significantly shorter than for sad sentences like “You open your email in-box on your birthday to find no new emails,” or “angry” sentences like “The workload from your pompous professor is unreasonable.” The authors tested and verified the prediction that paralysis of the corrugator supercilii (“frown muscle”) by Botulinum Toxin-A (botox) selectively hinders processing of angry and sad sentences, relative to happy sentences. Finding that BTX selectively slowed the reading of sentences that described situations that normally require the paralyzed muscle for expressing the emotions evoked by the sentences, they concluded that sentence comprehension involves a mental simulation of sentence content that calls on

685 686

693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741

Sentences “A nice line” is an often-used expression for phrases or sentences that, like the political slogan “I like Ike,” apply the two major principles of the poetic genre, i.e., the prominence of sound properties and more or less subtly expressed or perceived affective meanings, in an affectively and aesthetically appealing way (Jakobson, 1960; Schrott and Jacobs, 2011; Aryani et al., 2016). “What are men to rocks and mountains?” (Jane Austen, Pride and Prejudice) or “At the still point, there the dance is” (T. S. Eliot, “Four Quartets”) are examples for nice lines from novels and poetry. Fully explaining what makes such lines likeable or beautiful is beyond the powers of this paper. However, given our initial hypothesis that the computed valence of single words – and the mostly preconscious and preverbal EADs accompanying single word recognition – forms the basis of pleasurable, aesthetic literary experiences with more complex verbal materials, we can try to answer the simpler question: To what extent do the factors known to determine EADs to single words also co-determine the liking of sentences. The simplest model predicting the liking of a phrase or sentence should take into account only the affective meaning of its component words while neglecting other potentially relevant influences like word type, order, or syntactic role, or higher-level, stylistic features such novelty, originality, or metaphoricity (Jacobs, 2015a). If this null-model of supralexical affective meaning were correct, a simple declarative sentence containing a positive noun and a negative adjective like “The mother is bad” should – on average – be evaluated as neutral. As counter-intuitive as this may sound, the seminal studies in emotional stylometry by Bestgen (1994) and Whissell (1994) both demonstrated that the valence of supralexical units (sentences, texts) could be predicted – to a considerable extent – as a function of the valence of their component words. Correlating the valence ratings for words and sentences taken from four different texts (The Little Match Girl by Andersen, He Belonged to Me, Said the Sea by Cesbron, The Seven Ravens by the brothers Grimm, and The Stroll by de Maupassant), Bestgen (1994; Table 2) showed that word valence accounted for 30–60% of sentence valence, depending on the text. Even when considering potential effects of some methodological limitations regarding, e.g., the order of the ratings, these numbers are amazing. Intrigued by the success of the null-model, Lüdtke and Jacobs (2015) tested the hypothesis that the supralexical affective structure is a linear combination of the valences of the nouns and adjectives contained in simple declarative sentences without any claim of being “beautiful” or “poetic.” Examples for most liked sentences in that study are “Die Kellnerin ist dankbar” (the waitress is grateful) or “Der Kuss ist romantisch” (the kiss


742 Q12 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794

4

The higher arousal values for both negative adjectives and nouns than for positive ones likely played a role here. Since Bestgen (1994) did not measure/report arousal values, we can only speculate that apart from the lack of context, this is another potential source for discrepant results.

7


795 796 797 798

Jacobs et al.

799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855


the same neural systems used in literal action, perception, and emotion. Looking at more poetic materials than the simple declarative sentences, in a recent neuroimaging study, Citron and Goldberg (2014) found significantly enhanced activation of the left amygdala while participants read conventional, taste-related metaphors such as She looked at him sweetly compared with their literal counterparts, e.g., She looked at him kindly. Since the amygdala has been associated with automatic processing of highly emotionally arousing stimuli, as well as with salience or relevance detection, the authors suggested that metaphorical formulations per se may be more emotionally engaging than literal ones. This finding is in line with the results of two other studies already mentioned: Forgács et al.’s (2012) study on metaphorical and literal NNCs which also observed enhanced left amygdala activation for the former, and the meta-analysis of 23 neuroimaging studies on the comprehension of figurative statements (i.e., metaphors, idioms, proverbs, sarcastic, and ironic statements), in which activation of the left amygdala was reported for the contrast between figurative and literal statements (Bohrn et al., 2012a). Perhaps, given that amygdala activation is also associated with reward, if the amygdala is active in figurative language processing, it is being allowed to be so because the control regions of the prefrontal cortex-amygdala network linked to emotion regulation (e.g., Morawetz et al., 2016) detected positive affect. This might imply that metaphors and figurative language in general are, at least on an implicit neuronal level, more pleasant5 . To summarize, EADs to simple sentences can be related to enhanced meaning construction and processing dynamics based on the general principle of good things come easy (Unkelbach et al., 2010), which draws on Boucher and Osgood’s (1969) Pollyanna hypothesis (i.e., the universal human tendency to use positive words more frequently and diversely than negative ones in communicating) and on the mere exposure effect (Murphy and Zajonc, 1993). The structural argument in favor of a general processing advantage for positive stimuli, the informational density hypothesis6 (Unkelbach et al., 2008; Sylvester et al., 2016), is complemented by both a computational and phenomenological one: If positive stimuli are more densely clustered in memory, then people should experience higher subjective exposure to positive than to negative stimuli, because any positive stimulus has a greater likelihood of triggering thoughts of a large number of other associated positive stimuli (Hintzman, 1988; Hofmann and Jacobs, 2014). Thus, greater subjective exposure to (or familiarity with) positive words is an experiential consequence of the informational density hypothesis’ computational and structural arguments, as experimentally verified by Unkelbach et al. (2010). In synopsis with what was concluded for words and multiword expressions, the liking of sentences thus seems to depend on higher subjective exposure and ease of processing for positive

information. The participation of neural circuits for reward and relevance detection (medial orbitofrontal gyrus, caudate nucleus, amygdala) and, more generally, neural and bodily systems used in perception, action, and emotion fits with this interpretation, if we assume that under conditions of congruity they all contribute to the felt ease of processing and pleasure (Havas et al., 2010). From Kintsch’s (2012) computational perspective, the aesthetic liking or beauty of a nice line or sentence would be a complex function of harmony (i.e., minimal complexity), variety (i.e., distinctiveness of the parts), and compression (i.e., increasing harmony over processing time).

The authors thank review editor Matthew Pelowski for this interesting suggestion. 6 Positive information is more similar and is more closely related to other positive information than negative information is to other negative information, thus affording less precise mental representations and decisions/actions than negative information.


858 859 860 861 862 863 864 865 866 868

As we move from smaller and simpler to longer and more complex text materials, the likelihood of identifying the key features that determine EADs, both on the text and the reader side, will hardly increase7 . On the text side, one can analyze the potential structural key features via descriptive tools such as the 4 ⇥ 4 matrix that combines four text levels (metric, phonological, morpho-syntactic, semantic) with four groups of features (sublexical, lexical, interlexical, supralexical; Jacobs, 2015b). Many of these features can be quantified by appropriate tools like the BAWL, Coh-Metrix (Graesser et al., 2004) or TAACO (Crossley et al., 2015), and then fed into regression analyses to find out which features affect which response variables. On the reader side, longer pieces of text like stories and novels should increase the likelihood of triggering personal memories which play a key role for story comprehension (e.g., Larsen and Seilman, 1988; Pleh, 2003; Burke, 2011). It is beyond the scope of this paper to review even the most cited of the enormously broad and controversial literature on story processing (e.g., Wilensky, 1983, and commentaries to his paper), though. We therefore focus on a few aspects that seem most relevant for our computational modeling purposes (see Final section). There are numerous papers on both the cognitive and emotional components of story processing (e.g., Brewer and Lichtenstein, 1982). In his review, Bestgen (1994) concludes that to adopt an integrated approach of interest and storyness, as advocated by Stein (1982), structural models of story grammar and cognitive interest (e.g., Golden and Rumelhart, 1993) should be complemented by methods of text analysis allowing the quantification of the emotional content in stories (cf. Miall, 1989). Using the four stories mentioned above, Bestgen could show that the more his readers liked the words and sentences of a story, the more they also liked the story itself (R2 lin ranging from a minimum of 35% for the word-text valence correlation to a maximum of 70% for the sentence-text valence correlation for Maupassant’s story). Good “storytelling, inevitably, is about compelling human plights that are “accessible” to readers.” (Bruner, 1986, p. 35), i.e., make the comprehension of the plans underlying the goaldirected actions of its protagonists easy by facilitating situation model building and mental simulation through theory of mind

869

7

911

Here, we exclude micro-narratives like Hemingway’s famous “For Sale: Baby Shoes, Never Worn” which is based on what is left unsaid.

8

857

867

Stories

5

856


870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 912

Jacobs et al.

913 914 915 916 917 918 919 920 921 922 923 924 925

Q13

926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960

processes of empathy and identification which all facilitate immersion (e.g., Mason and Just, 2009; Mar, 2011; Altmann et al., 2012, 2014; Hsu et al., 2014; Jacobs and Lüdtke, 2017). According to Brewer and Lichtenstein (1982) stories differ from narratives in that they are structured to evoke a particular affective response pattern in the readers. These authors identified three generic patterns, suspense, surprise, and curiosity, which correspond to different discourse structures and encode a distinct functional operation of the mind within stories’ overall intersequencing, i.e., the dynamics of prospection, retrospection, and recognition, respectively (Sternberg, 2003; cf. Jacobs, 2015b). Simply put in terms of Brewer and Lichtenstein’s theory, a story will be liked if it succeeds in arousing and resolving readers’ affective responses, e.g., suspense (Jose and Brewer, 1984; see also Berlyne, 1971; Lehne et al., 2015). However, as has become clear from the preceding paragraphs, affective reader responses can be triggered at multiple levels of text by multiple features, from single words to the overall affective tone of the largest unit (e.g., a story). In addition, even though a story may have a dominant discourse structure of the type assumed by Brewer and Lichtenstein (1982), suspense, surprise, and curiosity responses can all happen in perhaps any kind of narrative thus making the theoretical liking function a more complex one than suggested by the former authors. Moreover, readers’ moods and motivations before choosing a text also are factors shaping their overall affective response, as are the sociocultural frames and actual contexts of the story or book and the proper reading act (e.g., knowledge about authors, reading motivation, etc.; Burke, 2011; Jacobs, 2011). Finally, while liking may be a compound but still relatively simple function of the affective reader responses in a certain type of literature, e.g., suspenseful crime stories, literary reading offers a wealth of other features that may trigger emotions and EADs. Thus, Miall (1989) was among the first to point out that defamiliarization or foregrounding (van Peer, 1986) also is an important source of affective reader responses and that readers of literary texts seek the experience of defamiliarization, i.e., look forward to redefine, modify, or suspend their schemata, and through this process explore the emotions of the self through engagement with the text. If Miall is correct, the latter process is the primary goal of reading. At the neuronal level, the liking of (short) stories and text passages has been associated with bilateral medial prefrontal cortex, supramarginal gyrus/temporoparietal junction, and left dorsolateral prefrontal cortex (Altmann et al., 2012), the left posterior middle temporal gyrus (Hsu et al., 2015), and the medial frontal cortex (Lehne et al., 20158 ).

961

Poetry

962

Poetry can be very short, like Pound’s two-liner “In a Station of the Metro” or Quasimodo’s “And suddenly it’s evening” (Jacobs, 2016a), or, even more extreme, the one-word micropoetry described in Limbach’s (2004) book on the most beautiful

963 964 965 966 967 968 969


German words of the year (Jacobs et al., 2015). It can also fill several pages like Schiller’s “Die Glocke” (The Song of the Bell, 430 lines) or a book like Homer’s Iliad. This is one reason why we deal with poems at the end of this section. Another reason is that poetry is often viewed as the highest or prototypical form of verbal art. A third reason is that poetry can generally be understood as inherently concerned with the expression and elicitation of emotions (Meyer-Sickendiek, 2011; Lüdtke et al., 2014) while being deeply rooted at the perceptual and aesthetic levels in the domains of speech and sound (Schrott and Jacobs, 2011). Richards (1929) pioneered the study of poetry reception by having about 100 literature students read a set of 13 poems deemed of different quality and write down their impressions. The results of his method of open-ended responses – plus the lack of any statistical analysis – lead him to conclude that he got a hundred different verdicts from a hundred readers. In a statistical reanalysis of Richards’ data and two empirical studies using his poetic materials – one using ratings, the other essay writing, Martindale and Dailey (1995) could show that rater agreement was significant on 37/40 scales thus contradicting Richards’ often repeated subjective conclusion. On the liking scale, the agreement was very high, but inopportunely Martindale and Dailey (1995) provide no data specifying which of the 13 poems was liked most and why. Trying to find quantitative structural predictors for the aesthetic success – a variable that is assumed to be highly correlated with liking – of the 154 Shakespeare sonnets, Simonton (1989, 1990) isolated four factors accounting for the superiority of very popular sonnets (1) They treat specific themes, e.g., friendship, beauty, love, or death. (2) They display considerable thematic richness in the number of issues discussed. (3) They exhibit greater linguistic complexity as gaged by quantitative measures like the type-token ratio (i.e., the ratio of different words to total words) or adjective-verb quotient (i.e., the proportion of adjectives to verbs). (4) They feature more primary process imagery (using the Regressive Imagery Dictionary/RID by Martindale, 1975).


972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012

9

1025

Generally, more highly esteemed products tend to treat a broader range of themes (Simonton, 1983).

9

971

Factor 1 was estimated by using the Syntopicon (Hutchins, 1952), a detailed topical index to the Great Books of the Western World. Overall, the 154 sonnets negotiate 24 different topics, some being variations on a key topic, such as intensity and power of love vs. altruistic love. Factor 2 basically is a count of the number of different topics within a sonnet9 . Factor 3 reflects lexical variability and verbal complexity including variables like the number of unique words, an index of novelty. Finally, factor 4 is derived from words that associate primary process imagery in the Freudian sense, e.g., orality, sex, anality, etc. In sum, according to Simonton’s computer-based analysis, the liking of Shakespeare sonnets can be related to a combination

8

The latter result is indirect in the sense that Lehne et al.’s parametric effects relate to suspense ratings which correlated very highly with liking (reading pleasure) ratings (r = 0.89), as argued, e.g., by Jose and Brewer (1984).

970


1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1026

Jacobs et al.

1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083


of the specificity-variability, novelty-complexity, and primary process imagery of the verbal material, all of which can be viewed to influence the arousal potential of a poem supposed to drive aesthetic pleasure (Berlyne, 1971; Cupchik, 1986). However, a recent more extensive and deeper computationallinguistic analysis of the 154 sonnets using a novel tool called SPARSAR challenges Simonton’s empirically not investigated deductions (Delmonte, 2016). Again, though, the computational data produced by SPARSAR remained empirically untested and will not be further discussed here. Focusing on lexico-semantic textual properties, the pioneering works of Martindale and Simonton appear to neglect a key feature of poetry: Its sound structure and euphony (e.g., Tsur, 1992). A recent example for an attempt to predict the liking of entire poems by quantifying the euphony-eusemy nexus typical for poetry (i.e., phonological iconicity; Jakobson and Waugh, 1979/2002) is given in Aryani et al. (2016). The authors used the EMOPHON software (Aryani et al., 2013) to obtain quantitative estimates of the basic affective tone of poems from Enzensberger’s (1957/1981) collection “verteidigung der wölfe” (in defense of the wolves). The 20% of variance in valence ratings accounted for in the 57 poems by this measure strongly suggests that the iconic properties of foregrounded phonological units are a co-determinant of liking. While having shown effects of phonological iconicity on poetry reception is no news per se, the quantification of the distinctive features and the fact that on top of liking ratings the EMOPHON software also allowed to predict specific emotional responses, such as “spitefulness ratings” (R2 lin = 0.22; see Table 2 in Aryani et al., 2016), makes the study a promising model for future approaches. Ullrich et al. (2016) offer an extended quantitative analysis of the reception of these 57 poems by Enzensberger, both on the text side and the reader response side. On the text side, based on the 4 ⇥ 4 matrix for textual analyses (Jacobs, 2015b), they added lexical and interlexical variables to the sublexical ones used in Aryani et al. to predict liking ratings. On the reader side, they added ratings for liking (in addition to valence ratings), poeticity, and onomatopoeia to those used by Aryani et al. For the present purposes, the most interesting results are: first, the strongest predictors of liking ratings were the valence and arousal values of the words in the poems, as computed using the BAWL (R2 = 0.47; p < 0.001). Although this is less variance accounted for than in the study by Bestgen (1994) using stories, it is worth noting that – at least in this extensive sample of German poetic texts – the affective-semantics of single words explain about half of the “liking cake.” Second, across all 57 poems, valence, and liking ratings were only moderately correlated (r = 0.59, R2 = 0.35; all p < 0.001). This value varied considerably for individual poems and poem categories, though. Thus, for sad poems, valence predicted liking better (R2 = 0.45; p < 0.0001) than for the two other groups: spiteful (R2 = 0.29; p < 0.025) and friendly (R2 = 0.23; p < 0.039). These variable, small to moderate, relationships between two reader response measures suggest that although ratings for valence, liking, or beauty seem to have some common latent factor (Lüdtke et al., 2014), they partially tap into different processes.


A third recent model-guided, multimethod study (Jacobs et al., 2016) used a mixture of tools for predicting aesthetic liking ratings, as well as other dependent variables, such as electrodermal activity, heart rate, or responses to the Poetry Reception Questionnaire for 24 German poems constituting the “mood poetry” corpus (Lüdtke et al., 2014). Similarly to Ullrich et al. (2016), it quantified their distinctive features based on the 4 ⇥ 4 matrix and other text-analytic tools, but added a supralexical level: the motif group of the poems (Stillness, Space, City, and Morning). Here, the variance accounted for in liking ratings is of interest: supralexical factor motif (R2 = 0.39, p < 0.017), interlexical factors valence span, and arousal span (R2 = 0.14, p < 0.018 and 0.13, p < 0.019, respectively). Furthermore, one sublexical (phonological iconicity) and 10 lexical variables significantly affected liking ratings (Table 1), showing a degree of complexity comparable to that of the single word VDT study by Jacobs et al. (2015). Neuronal correlates of processing (not of liking) poetic (vs. non-poetic) texts are the bilateral precentral and inferior frontal gyrus, as well as the right dorsolateral prefrontal cortex extending into the anterior insula, and beyond to the temporal pole. Interestingly, the dorsomedial prefrontal cortex showed reduced activation during reading of poetic pieces, compared to the reading of prosaic pieces (O’Sullivan et al., 2015). Further areas specifically related to poetry reception are the right posterior/mid-cingulate, parahippocampal, and left superior temporal gyrus, as well as bilateral hippocampus (Zeman et al., 2013). To summarize, as hypothesized by the NCPM, liking decisions concerning poetry are affected by a multitude of sublexical, lexical, interlexical, and supralexical factors at all four text levels considered by Jakobson (metric, phonological, morpho-syntactic, and semantic). Their dynamic interactions make it difficult to obtain a clear picture about the relative weight each factor has in determining EADs. Without proper process models (e.g., Leder et al., 2004, for visual arts; Jacobs, 2011, 2015a; for literary reading) or computational models (e.g., Hofmann and Jacobs, 2014, for word recognition) offering testable predictions at complementing levels of observation (i.e., neuronal, experiential, behavioral), the hidden structure and dynamics underlying EADs and aesthetic liking for complex verbal materials will be hard to uncover.

1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126

SUMMARY OF PREVIOUS SECTIONS

1127 1128 1129

To facilitate integration of the previous sections, Table 1 summarizes, in a very simplified way, the key pieces of information relevant for EADs to verbal materials of increasing complexity. The data in the various cells of this Table are hard to compare or integrate into a meaningful picture, though, because the materials, tasks, and methods used in the studies producing them were simply too heterogeneous. What would be needed for an integrative account of EADs to different verbal materials are studies comparing, say, literal and figurative sentences, story passages or poems where all materials are analyzed with the same text-analytical tools (to identify quantitative predictors)

10


1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140

Jacobs et al.

1141 1142 1143 1144

TABLE 1 | Stimulus features and neural networks relevant for liking verbal materials.

1147 1148

INSERT

1149 1150

Relevant neural networks1

Words

valence, arousal, imageability, frequency/familiarity, number of syllables, neighborhood density, joy/happiness, fear, anger, sadness, disgust, taste, grasp, move

orbitofrontal and insular cortex, middle temporal and left inferior frontal gyrus, hippocampus

Idioms

familiarity, arousal, figurativeness

left pre- and post-central gyri, right superior temporal gyrus

(Anti-)Proverbs

familiarity, rhyme, meter, prägnanz

medial orbitofrontal cortex, striatum

Sentences

valence congruity, figurativeness, prägnanz ,harmony, variety, compression left amygdala, medial orbitofrontal gyrus, caudate nucleus

Stories

valence, arousal, features evoking suspense, surprise and curiosity

bilateral medial prefrontal cortex, supramarginal gyrus/temporoparietal junction, left dorsolateral prefrontal cortex, left posterior middle temporal gyrus

Poems

basic affective tone (phonological iconicity), word valence and arousal, imageability, taste, grasp, move, joy, fear, anger, sadness, disgust, valence and arousal span, motif, specificity, thematic richness, linguistic complexity, primary process imagery

bilateral precentral and inferior frontal gyrus, right dorsolateral prefrontal gyrus, anterior insula, temporal pole, posterior/mid-cingulate, parahippocampal and left superior temporal gyrus, bilateral hippocampus

1209

networks specifically related to (dis-)liking are mentioned excluding all other structures related to, e.g., visual-auditory language processing. The exception are poems for which the cited studies provided no data about liking; here the relevant networks refer to those that are recruited during poetry vs. prose reading.

1213

1153 1154 1155 1156 1157

1199

Relevant features

1151 1152

1198

Verbal stimulus

1145 1146


1200

1 Only

1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171

1174 1175 1176

mining/machine learning technique – to illustrate how EADs to high-dimensional stimuli such as words (or text segments) can be predicted. A potent use of decision tree modeling is exploring relationships without having a good prior theoretical model: It can handle even large data problems efficiently allowing to test clear hypotheses, and the results are usually transparent and easily interpretable10 . Here we were interested in asking the question which of 10 quantitative features of words related to their affective-semantics and empirically established as potentially relevant factors was most important in determining EADs.

1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197

1203 1204 1205 1206 1207 1208 1210 1211 1212 1214

DECISION TREE (RECURSIVE PARTITIONING) MODELING OF EADS TO SINGLE WORDS As stated above, the number of factors involved in determining EADs to longer and more complex text materials such as stories, poems, or novels represents a huge challenge to any modeling approach. First descriptive models have taken up this challenge only recently, e.g., the NCPM (Jacobs, 2011, 2015a, 2016b) or Burke’s (2011) model of reading novels, but the way to formal computational models has yet to be paved by further research sufficiently constraining theory development (for a review of model types, see Jacobs and Grainger, 1994 or Hofmann and Jacobs, 2014). Note that even EADs to simple verbal materials such as single words still have not been modeled computationally, despite a wealth of formal models in the field of decision theory and neuroeconomics (e.g., Rangel and Clithero, 2014). One reason is that performance in the VDT, i.e., RTs reflecting positive vs. negative decisions, may depend on more than a dozen quantifiable factors. To prepare development of more formal modeling tools, we recently have proposed three descriptive neurocognitive models of the VDT (Jacobs et al., 2015). In this section, we go a step further by adopting an exemplary, formal decision tree modeling approach – a standard data


1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228

1177 1178

1202

1215

and in which the same task and method is used (e.g., liking ratings, VDT, fMRI). Meanwhile, Table 1 can serve as a first orienting tool for future studies in Neurocognitive Poetics and affective decision making informing researchers about which factors they should carefully consider using as independent or controlled/matched variables in their designs. An important task for future studies is to determine the relative weight and potential dynamic interactions of these features. Moreover, column three of Table 1 could help to select regions of interest or seed regions for neuroimaging studies using psychophysiological interaction (PPI) or dynamic causal modeling (DCM) analyses (e.g., Altmann et al., 2012; Lehne et al., 2015; Pehrs et al., 2015).

1172 1173

1201

STEPWISE MODELING APPROACH

1229

Similar to Jacobs et al’s. (2015) statistical analyses, we used a stepwise approach going from simple to complex models to see how much complexity in the input space is necessary to obtain an adequate model performance. All models were trained on 70% of the words (the randomly chosen training set is the part that estimates model parameters) and then validated on the remaining 30% (the validation set is the part that addresses or validates the predictive ability of the model). The stimuli were the N = 91 (46 negative and 45 positive) words used in Jacobs et al.’s (2015) reanalysis of the original BAWL study for which a large number of quantifiable features are available. We started with a simple two-variable “affective-semantic” model that implemented the hypothesis that EADs to words depend on their arousal and imageability values, as quantified in the BAWL database (cf. Jacobs et al., 2015). The more complicated threevariable “embodiment” model tests the hypothesis that EADs mainly depend on word features related to embodied associations (taste, grasp, move). The next five-variable model posits that the associations with five discrete “basic” emotions (Joy, Fear, Anger, Sadness, Disgust) determine EADs. Finally the most complex

1231

10

A classic application is when you want to turn a data table of symptoms and diagnoses of a certain illness into a hierarchy of questions. These questions help diagnose new patients more quickly.

11


1230 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254

Jacobs et al.

1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298


model included all 10 variables of the previous models. The models were implemented using the PARTITION tool of the JMP Pro 11 software and model performance was gaged by the number of correct EADs, i.e., whether the model classified a target word correctly as either positive or negative (corresponding to the norm values of the BAWL database). Descriptively, model performance is expressed by the number of partitions, i.e., how many decisions are required to obtain maximum accuracy, R2 and the number of misclassifications, i.e., how often the model classified a positive stimulus as negative or vice versa. Table 2 summarizes the results. Each model in the table implements and tests a different hypothesis concerning the factors determining EADs to single words, e.g., Model 1 tests to what extent EADs are driven by two affective-semantic features only. Descriptively, models 3 and 4 are the winners of this competition both producing an almost perfect performance with only one misclassification, i.e., 99% correct responses in valence decisions to single words. Thus, somewhat surprisingly, the data in Table 2 suggest that the most complex model with 10 input features is not necessarily the best predictor of EADs, since the simpler five-variable discrete emotion model fared as well. Can we tentatively infer that EADs to words are determined by their associations to five basic emotions then (Westbury et al., 2014)? Figure 1 illustrating the decision tree data for Model 3 suggests an even simpler answer: to obtain 99% correct EADs, the model requires only two questions. Question 1 is whether the JOY/HAPPINESS value of a word is bigger than or equal to 1.9 (on the five point scale of the BAWL). If the answer is “No” the EAD will be “Negative” (0 in this case; red circles). Question 2 then applies to words for which the answer is “Yes” (1; blue circles) and is whether their DISGUST value is smaller than 1.5. If it is, the EAD will be positive, i.e., the word will get a “thumb up” or LIKE. The results illustrated in Figure 1 tempt a straightforward interpretation: what basically determines EADs to words from our empirically well cross-validated BAWL sample is the degree to which they are associated with joy or happiness. To a lesser extent, disgust associations also play a role. Associations with the other three discrete emotions (sadness, anger, fear), however, are not really useful for EADs in this context. That joy apparently plays a much stronger role in EADs as simulated by our Model 3 fits well with the positivity superiority effect mentioned above (Lüdtke and Jacobs, 2015) which can be interpreted by the

hypothesis that the hippocampus is more generally involved in the processing of positive affect (Hofmann and Kuchinke, 2015). Given the limitations of this heuristic modeling approach (a rather small sample of 91 German stimuli; only single words), any interpretation can only be tentative, though. Nevertheless, these results fit well with the evolutionary and appraisal accounts of aesthetic emotions discussed in the “Introduction” and also with data from recent neurocognitive studies using either lexical or VDTs (Briesemeister et al., 2014, 2015; Kuhlmann et al., 2016) and thus are further (computational) evidence for our specific hypothesis stated in the “Introduction.” Together these results suggest that valence is indeed a compound superfeature neuronally computed at the so-called tertiary (i.e., neocortical) level of affective processing according to Panksepp’s (1998) hierarchical theory of emotions. In contrast, joy/happiness and disgust are more basic and central affective responses likely computed at the secondary level (i.e., the limbic system). The neuroimaging results from Briesemeister et al. (2015) indicate that words associated with joy produce reduced brain activity in the amygdala, i.e., at the secondary level of Panksepp’s theory, while words that have positive valence, but are not associated with the basic emotion joy/happiness activate the orbitofrontal cortex at the tertiary level of affective processing. Further evidence for this comes from a brain-electrical experiment using the same stimuli and indicating that joy-words affect the early N1 component of the ERP – known to be sensitive to affective conditioning (Fritsch and Kuchinke, 2013) –, while positively valenced words affect the later N400 and LPC responses. Additional evidence stems from the above mentioned study by Ponz et al. (2013) showing that the anterior insula – which is also considered part of the secondary level – is activated during the reading of disgusting words. The authors interpreted this finding in terms of neural reuse and the Panksepp-Jakobson hypothesis (see “Introduction”) suggesting that phylogenetically younger processes such as reading rely at least partially on already-existing ancient affective circuits like the limbic system. Words strongly associated with a given emotion are thus assumed to activate the corresponding conditioned affect program more strongly than words judged as being weakly related to that emotion, or than neutral words. To summarize, the results from our decision tree modeling together with those from recent neurocognitive studies allow to simplify our initial main hypothesis for straightforward testing in future studies on EADs and aesthetic liking: The extent to which high-dimensional stimuli such as words are associated with two basic emotions – likely to be evoked at the secondary level in Panksepp’s (1998) emotion theory –, namely joy/happiness and disgust, drives EADs in a considerable and quantitatively predictable way. Whether this is the case only for single words (for which the necessary feature values can be found in databases like the BAWL) or whether it can be generalized to more complex verbal and non-verbal materials is an open issue for future studies. Note that according to this simplified hypothesis the role played by associations to other basic emotions (e.g., fear, sadness, anger) or by other affective and embodied features (e.g., arousal, taste) seems to be a relatively minor one. The empirical data of the study by Jacobs et al. (2016) would suggest that with

1299 1300 1301 1302 1303 1304

TABLE 2 | Input variables and performance evaluation for four decision tree models of the valence decision task (VDT) with single words. Model

Input variables

1305 1306

Model performance (Nbr of partitions, R2 , misclassifications)

1307

(1) Affective-semantic

Arousal, imageability

6,.79, 6

1308

(2) Embodiment

Taste, grasp, move

12,.43, 18

1309

(3) Discrete

Joy, fear, anger, sadness, disgust

2,.94, 1

(4) All 10 features

All above features

2,.94, 1

1310 1311


12


1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368

Jacobs et al.

Q4 Q5


1369

1426

1370

1427

1371

1428

1372

1429

1373

1430

1374

1431

1375

1432

1376

1433

1377

1434

1378

1435

1379

1436

1380

1437

1381

1438

1382

1439

1383

1440

1384

1441

1385

1442

1386

1443

1387

1444

1388

1445

1389

1446

1390

1447

1391

1448

1392

1449

1393

1450

1394

1451

1395

1452

1396

1453

1397

1454

1398

1455

1399

1456

1400

1457

1401

1458

1402

1459

1403

1460

1404

1461

1405

1462

1406

1463

1407

1464

1408

1465

1409 1410 1411

FIGURE 1 | Upper panel. Partitioning results and decision tree report for 46 negative (red) and 45 positive (blue) words as implemented in Model 3. Resp, response (positive vs. negative; see text for details). Lower panel. Detailed decision tree for Model 3 with number of candidates, G2 values indicating the likelihood ratio $2 for the best split, and LogWorth statistics [defined as – log10(p-value)]. The optimal split is the one that maximizes the LogWorth.

1412 1413 1414 1415 1416 1417 1418

1421

complex materials such as poems also associations with these other features gain in weight, but implementing the 24 poems studied by these authors in a decision tree or other formal model clearly is beyond the scope of this paper and will thus await future research.

our group, we submitted the hypothesis that the main factor used in the investigation of EADs, i.e., stimulus valence, is a semantic superfeature resulting from a yet unknown integration of experiential and distributional data, at least partially represented in associative activation patterns of affective-semantic networks starting out in parts of the limbic system (Briesemeister et al., 2014, 2015; Jacobs et al., 2015; Kuhlmann et al., 2016; Ponz et al., 2016). 2013 Our review of methods and materials used in the scientific study of EADs, as well as of the factors determining the liking of verbal materials with increasing degrees of complexity provided accumulated empirical evidence for this assumption.

CONCLUSION AND OUTLOOK

1422 1423 1424 1425

1467 1468 1469

1419 1420

1466

We started out with Virginia Woolf ’s question How did one add up this and that and conclude that it is liking one felt, or disliking? Based on previous empirical and theoretical work from


13


1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482

Jacobs et al.

1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508


A final section presenting formal decision tree models of EADs with different degrees of complexity (2–10 input variables) also supported the hypothesis and suggested a most parsimonious version, i.e., a compression to only two relevant, unevenly weighted dimensions of the semantic space supporting EADs to single words: joy/happiness and disgust. Thus, to use Woolf ’s words, what is added up for concluding a LIKE or DISLIKE decision, at the neuronal level probably relates to two ancient core affect programs, well described in Panksepp’s (1998) emotion theory. Liking and beauty ratings may be a complex context-dependent function of many variables, and so may be EADs (Güçlütürk et al., 2016; Marin et al., 2016), but perhaps in the end it always involves implicit associations with joyful or disgusting events triggered by the stimulus, the former taking a greater weight in the decision. Future studies on EADs should thus pay more attention to the underlying (asymmetric) affective-semantic structure (i.e., relating to discrete or embodied emotions) of the stimuli used, before a priori polarizing them symmetrically into “positive” or “negative” ones. Studies in neuroeconomics focusing on the reward values of stimuli and actions (e.g., Rangel and Clithero, 2014) should pay more attention to the idea that although EADs entail no direct overt reward, they may well be guided by unevenly weighted intrinsic rewards based on remembrance of things past. This idea, of course, is not new. For example, Epstein’s (2004) neuroaesthetics theory, inspired by Proust and James,

claims that the function of art is to evoke the underlying associative network indirectly in the mind of the observer by using carefully chosen sensory surfaces to control the stream of thought and induce pleasure and aesthetic liking. This involves distinct neural/cognitive mechanisms, including a network of associations supported largely by the medial temporal lobes (e.g., hippocampus) that determines the relationship between the current nucleus and other potential thoughts and feelings forming the fringe (for a more detailed treatment, cf. Jacobs, 2015b). The challenge is to uncover these mechanisms in all their apparent complexity by combining neurocognitive studies and computational modeling, as tentatively exemplified in this paper. A final word, lent from Kintsch (2012, p. 646), on the liking of verbal stimuli that were left out from the present paper is in order: books. “Why is it that we can read a great book many times and it becomes more interesting with each reading? Because it affords us the opportunity to fine-tune our (mental) model, to construct a novel interpretation every time. The book remains the same, but we – our model – change.”

1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538

1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560

AUTHOR CONTRIBUTIONS

1561 1562 1563

AJ: Paper conception and major writing. MH and AK: minor writing.

1509 1510

1540

1564 Q8 1565 1566 1567

REFERENCES

Bohrn, I. C., Altmann, U., Lubrich, O., Menninghaus, W., and Jacobs, A. M. (2012b). Old proverbs in new skins—an FMRI study on defamiliarization. Front. Psychol. 3:204. doi: 10.3389/fpsyg.2012.00204 Bohrn, I. C., Altmann, U., Lubrich, O., Menninghaus, W., and Jacobs, A. M. (2013). When we like what we know—a parametric fMRI analysis of beauty and familiarity. Brain Lang. 124, 1–8. doi: 10.1016/j.bandl.2012.10.003 Boucher, J., and Osgood, C. E. (1969). The Pollyanna hypothesis. J. Verbal Learning Verbal Behav. 8, 1–8. doi: 10.1016/S0022-5371(69)80002-2 Boyd, B. (2001). The origin of stories: horton hears a who. Philos. Literature 25.2, 197–214. doi: 10.1353/phl.2001.0022 Brewer, W. F., and Lichtenstein, E. H. (1982). Stories are to entertain: a structural-affect theory of stories. J. Pragmat. 6, 473–486. doi: 10.1016/03782166(82)90021-2 Briesemeister, B. B., Kuchinke, L., and Jacobs, A. M. (2011). Discrete emotion norms for nouns – Berlin Affective Word List (DENN-BAWL). Behav. Res. Methods 43, 441–448. doi: 10.3758/s13428-011-0059-y Briesemeister, B. B., Kuchinke, L., and Jacobs, A. M. (2012). Emotional valence–A bipolar continuum or two independent dimensions? SAGE Open 2, 1–12. doi: 10.1177/2158244012466558 Briesemeister, B. B., Kuchinke, L., and Jacobs, A. M. (2014). Emotion word recognition: discrete information effects first, continuous later? Brain Res. 1564, 62–71. doi: 10.1016/j.brainres.2014.03.045 Briesemeister, B. B., Kuchinke, L., Jacobs, A. M., and Braun, M. (2015). Emotions in reading: dissociation of happiness and positivity. Cogn. Affect. Behav. Neurosci. 15, 287–298. doi: 10.3758/s13415-014-0327-2 Brown, S., Gao, X., Tisdelle, L., Eickhoff, S. B., and Liotti, M. (2011). Naturalizing aesthetics: brain areas for aesthetic appraisal across sensory modalities. Neuroimage 58, 250–258. doi: 10.1016/j.neuroimage.2011.06.012 Bruner, J. (1986). Actual Minds, Possible Worlds. Cambridge, MA: Harvard University Press. Bühler, K. (1934). Sprachtheorie: Die Darstellungsfunktion der Sprache [Language Theory: The Representational Function of Language]. Jena: G. Fischer. Burke, M. (2011). Literary Reading, Cognition and Emotion: An Exploration of the Oceanic Mind. London: Routledge.

Altmann, U., Bohrn, I. C., Lubrich, O., Menninghaus, W., and Jacobs, A. M. (2012). The power of emotional valence-from cognitive to affective processes in reading. Front. Hum. Neurosci. 6:192. doi: 10.3389/fnhum.2012.00192 Altmann, U., Bohrn, I. C., Lubrich, O., Menninghaus, W., and Jacobs, A. M. (2014). Fact vs fiction-how paratextual information shapes our reading processes. Soc. Cogn. Affect. Neurosci. 9, 22–29. doi: 10.1093/scan/nss098 Andrews, M., Vigliocco, G., and Vinson, D. (2009). Integrating experiential and distributional data to learn semantic representations. Psychol. Rev. 116, 463– 498. doi: 10.1037/a0016261 Anz, T. (1998). Literatur und Lust: Glück und Unglück beim Lesen. Munich: C.H. Beck. Aryani, A., Jacobs, A. M., and Conrad, M. (2013). Extracting salient sublexical units from written texts: “Emophon,” a corpus-based approach to phonological iconicity. Front. Psychol. 4:654. doi: 10.3389/fpsyg.2013.00654 Aryani, A., Kraxenberger, M., Ullrich, S., Jacobs, A. M., and Conrad, M. (2016). Measuring the basic a ective tone of poems via phonological saliency and iconicity. Psychol. Aesthet. Creat. Arts 10, 191–204. doi: 10.1037/aca000 0033 Ashby, F. G., and Isen, A. M. (1999). A neuropsychological theory of positive affect and its influence on cognition. Psychol. Rev. 106, 529–550. doi: 10.1037/0033295X.106.3.529 Bar, M., Kassam, K. S., Ghuman, A. S., Boshyan, J., Schmid, A. M., Dale, A. M., et al. (2006). Top-down facilitation of visual recognition. Proc. Natl. Acad. Sci. U.S.A. 103, 449–454. doi: 10.1073/pnas.0507062103 Berlyne, D. E. (1971). Aesthetics and Psychobiology. New York, NY: AppletonCentury-Crofts. Bestgen, Y. (1994). Can emotional valence in stories be determined from words? Cogn. Emot. 8, 21–36. doi: 10.1080/02699939408408926 Bohrn, I. C., Altmann, U., and Jacobs, A. M. (2012a). Looking at the brains behind figurative language—A quantitative meta-analysis of neuroimaging studies on metaphor, idiom and irony processing. Neuropsychologia 50, 2669–2683. doi: 10.1016/j.neuropsychologia.2012.07.021

1539

1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596


14


Jacobs et al.

1597 1598 1599 1600 1601 1602 1603 1604 1605 1606

Q14

1607 1608 1609 1610

Q15

1611

2015 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652


Burke, M. (2013). The rhetorical neuroscience of style: on the primacy of style elements during literary discourse processing. J. Lit. Semantics 42, 199–216. doi: 10.1515/jls-2013-0010? Burke, M. (2015). The neuroaesthetics of prose fiction: pitfalls, parameters and prospects. Front. Hum. Neurosci. 9:442. doi: 10.3389/fnhum.2015.00442 Cacciari, C. (2014). Processing multiword idiomatic strings: many words in one? Ment. Lex. 9:267–293. doi 10.1075/ml.9.2.05cac doi: 10.1075/ml.9.2.05cac Chatterjee, A., and Vartanian, O. (2014). Neuroaesthetics. Trends Cogn. Sci. 18, 370–375. doi: 10.1016/j.tics.2014.03.003 Citron, F. M. (2012). Neural correlates of written emotion word processing: a review of recent electrophysiological and hemodynamic neuroimaging studies. Brain Lang. 122, 211–226. doi: 10.1016/j.bandl.2011 Citron, F. M., Cacciari, C., Funke, J. M., Hsu, C. T., and Jacobs, A. M. (2016a). Idiomatic expressions elicit stronger emotional responses and enhanced processing than literal sentences, even when their content is emotionally neutral. In revision Citron, F. M., Cacciari, C., Kucharski, M., Beck, L., Conrad, M., and Jacobs, A. M. (2016b). When emotions are expressed figuratively: psycholinguistic and affective norms of 619 idioms for German (PANIG). Behav. Res. Methods 48, 91–111. doi: 10.3758/s13428-015-0581-4 Citron, F. M. M., and Goldberg, A. E. (2014). Metaphorical sentences are more emotionally engaging than their literal counterparts. J. Cogn. Neurosci. 26, 2585–2595. doi: 10.1162/jocn_a_00654 Crossley, S. A., Kyle, K., and McNamara, D. S. (2015). The tool for the automatic analysis of text cohesion (TAACO): automatic assessment of local, global, and text cohesion. Behav. Res. Methods doi: 10.3758/s13428-015-0651-7 [Epub ahead of print]. Cupchik, G. C. (1986). A decade after Berlyne: new directions in experimental aesthetics. Poetics 15, 345–369. doi: 10.1016/0304-422X(86)90003-3 Delmonte, R. (2016). Exploring Shakespeare’s Sonnets with SPARSAR. Linguist. Literature Stud. 4, 61–95. doi: 10.13189/lls.2016.040110 Enzensberger, H. M. (1957/1981). Verteidigung der Wölfe: Gedichte. Frankfurt: Suhrkamp. Epstein, R. (2004). Consciousness, art and the brain: lessons from Marcel Proust. Conscious. Cogn. 13, 213–240. doi: 10.1016/s1053-8100(03)00006-0 Eysenck, H. J. (1941). The empirical determination of an aesthetic formula. Psychol. Rev. 48, 83–92. doi: 10.1037/h0062483? Forgács, B., Bohrn, I. C., Baudewig, J., Hofmann, M. J., Pléh, C., and Jacobs, A. M. (2012). Neural correlates of combinatorial semantic processing of literal and figurative noun-noun compound words. Neuroimage 63, 1432–1442. doi: 10.1016/j.neuroimage.2012.07.029 Freud, S. (1907). Der Dichter und das Phantasieren. Wien: Internationaler Psychoanalytischer Verlag. Fritsch, N., and Kuchinke, L. (2013). Acquired affective associations induce emotion effects in word recognition: an ERP study. Brain Lang. 124, 75–83. doi: 10.1016/j.bandl.2012.12.001 Golden, R. M., and Rumelhart, D. E. (1993). A parallel distributed processing model of story comprehension and recall. Discourse Process 16, 203–237. doi: 10.1080/01638539309544839 Graesser, A. C., McNamara, D. S., Louwerse, M. M., and Cai, Z. (2004). CohMetrix: analysis of text on cohesion and language. Behav. Res. Methods Instrum. Comput. 36, 193–202. doi: 10.3758/BF03195564 Graf, R., Nagler, M., and Jacobs, A. M. (2005). Factor analysis of 57 variables in visual word recognition. Z. Psychol. 213, 205–218. doi: 10.1026/00443409.213.4.205 Güçlütürk, Y., Jacobs, R. H. A. H., and van Lier, R. (2016). Liking versus Complexity: decomposing the Inverted U-curve. Front. Hum. Neurosci. 10:112. doi: 10.3389/fnhum.2016.00112 Harris, Z. S. (1951). Methods in Structural Linguistics, Vol. XV. Chicago, IL: University of Chicago Press. Havas, M., Gutowski, K. A., Lucarelli, M. J., Davidson, R. J., Havas, D. A., and Glenberg, A. (2010). Cosmetic use of Botulinum Toxin-A affects processing of emotional language. Psychol. Sci. 21, 895–900. doi: 10.1177/0956797610374742 Herbert, C., Ethofer, T., Anders, S., Junghöfer, M., Wildgruber, D., Grodd, W., et al. (2009). Amygdala activation during reading of emotional adjectives – An advantage for pleasant content. Soc. Cogn. Affect. Neurosci. 4, 35–49. doi: 10.1093/scan/nsn027

Hernadi, P., and Steen, F. (1999). The tropical landscapes of proverbia: a crossdisciplinary travelogue. Style 33, 1–20. Herrington, J. D., Mohanty, A., Koven, N. S., Fisher, J. E., Stewart, J. L., Banich, M. T., et al. (2005). Emotion-modulated performance and activity in left dorsolateral prefrontal cortex. Emotion 5, 200–207.? doi: 10.1037/15283542.5.2.200 Hill, A. B., and Kemp-Wheeler, S. M. (1989). The influence of context on lexical decision times for emotionally aversive words. Curr. Psychol. 8, 219–227. doi: 10.1007/BF02686750 Hintzman, D. (1988). Judgments of frequency and recognition memory in a multiple-trace memory model. Psychol. Rev. 95, 528–551. doi: 10.1037/0033295X.95.4.528 Hofmann, M. J., Dambacher, M., Jacobs, A., Kliegl, R., Radach, R., Kuchinke, L., et al. (2014). Occipital and orbitofrontal hemodynamics during naturally paced reading: an fNIRS study. Neuroimage 94, 193–202. doi: 10.1016/j.neuroimage.2014.03.014 Hofmann, M. J., and Jacobs, A. M. (2014). Interactive activation and competition models and semantic context: from behavioral to brain data. Neurosci. Biobehav. Rev. 46, 85–104. doi: 10.1016/j.neubiorev.2014. 06.011 Hofmann, M. J., and Kuchinke, L. (2015). “Anything is good that stimulates thought” in the hippocampus Comment on “The quartet theory of human emotions: an integrative and neurofunctional model” by S. Koelsch et al. Phys. Life Rev. 13, 58–60. doi: 10.1016/j.plrev.2015.04.007 Hsu, C.-T., Conrad, M., and Jacobs, A. M. (2014). Fiction feelings in Harry Potter: haemodynamic response in the mid-cingulate cortex correlates with immersive reading experience. Neuroreport 25, 1356–1361. doi: 10.1097/WNR.0000000000000272 Hsu, C. T., Jacobs, A. M., Citron, F., and Conrad, M. (2015). The emotion potential of words and passages in reading Harry Potter – An fMRI study. Brain Lang. 142, 96–114. doi: 10.1016/j.bandl.2015.01.011 Hutchins, R. M. (ed.) (1952). Great Books of the Western World: Syntopicon, 2 Vols. Chicago, IL: University of Chicago Press. Hyde, H. (1962). The Trials of Oscar Wilde. New York, NY: Dover Publications. Jacobs, A. (2015c). The scientific study of literary experience: sampling the state of the art. Sci. Study Lit. 5, 139–170. doi: 10.1075/ssol.5.2.01jac Jacobs, A. (2016a). The scientific study of literary experience and neuro-behavioral responses to Literature: reply to commentaries. Sci. Study Lit. 6. 164-174. doi Jacobs, A. M., and Lüdtke, J. (2017). “Immersion into narrative and poetic worlds: a neurocognitive poetics perspective,” in The Handbook of Narrative Absorption, eds M. Kuipers and F. Hakemulder (Amsterdam: John Benjamins). Jacobs, A. M. (2011). “Neurokognitive poetik: elemente eines modells des literarischen lesens [Neurocognitive poetics: elements of a model of literary reading],” in Gehirn und Gedicht: Wie wir unsere Wirklichkeiten Konstruieren [Brain and Poetry: How We Construct Our Realities], eds R. Schrott and A. M. Jacobs (Munich: Carl Hanser), 492–520. Jacobs, A. M. (2015a). “Towards a neurocognitive poetics model of literary reading,” in Towards a Cognitive Neuroscience of Natural Language Use, ed. R. Willems (Cambridge: Cambridge University Press), 135–159. doi: 10.1017/CBO9781107323667.007 Jacobs, A. M. (2016b). “Affective and aesthetic processes in literary reading: a neurocognitive poetics perspective,” in Dialogues between Literature and Cognition, eds M. Burke and E. Troscianko (Oxford: Oxford University Press). Jacobs, A. M., and Grainger, J. (1994). Models of visual word recognition: sampling the state of the art. J. Exp. Psychol. Hum. 20, 1311–1334. doi: 10.3758/s13428012-0236-7 Jacobs, A. M., and Kinder, A. (2015). “Worte als Worte erfahren: wie erarbeitet das Gehirn Gedichte (Experience words as words: how the brain constructs poems),” in Kind und Gedicht (Child and Poem), ed. A. Pompe (Berlin: Rombach), 57–76. Jacobs, A. M., Lüdtke, J., Aryani, A., Meyer-Sickendiek, B., and Conrad, M. (2016). Mood-empathic and aesthetic responses in poetry reception: a model-guided, multilevel, multimethod approach. Sci. Study Lit. 6. Jacobs, A. M., and Schrott, R. (2015). Gefesselt im Kopfkino: Von Kippschaltern, Madeleine Effekten und Don Quichote Syndromen bei der Immersion in Textwelten (Captivated in the Mind’s Cinema: Of Trigger-Switches, Don Quichote Syndroms and Immersion in Text Worlds). Available at: FIKTION.CC

1653

1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683

1684 Q17

10.1075/s 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710


15


Q16

Jacobs et al.

1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766


Jacobs, A. M. (2015b). Neurocognitive poetics: methods and models for investigating the neuronal and cognitive-affective bases of literature reception. Front. Hum. Neurosci. 9:186. doi: 10.3389/fnhum.2015.00186 Jacobs, A. M., Võ, M. L., Briesemeister, B. B., Conrad, M., Hofmann, M. J., Kuchinke, L., et al. (2015). 10 years of BAWLing into affective and aesthetic processes in reading: what are the echoes? Front. Psychol. 6:714. doi: 10.3389/fpsyg.2015.00714 Jacobsen, T. (2006). Bridging the arts and sciences: a framework for the psychology of aesthetics. Leonardo 39, 155–162. doi: 10.1162/leon.2006.39.2.155 Jakobson, R. (1944). Kindersprache, Aphasie und Allgemeine Lautgesetze (Reprint 1969). Frankfurt: Suhrkamp. Jakobson, R. (1960). “Closing statement: linguistics and poetics,” in Style in Language, ed. T. A. Sebeok (Cambridge, MA: MIT Press), 350–377. Jakobson, R., and Waugh, L. R. (1979/2002). The Sound Shape of Language. Berlin: Walter de Gruyter. doi: 10.1515/9783110889451 Jose, P. E., and Brewer, W. F. (1984). Development of story liking: character identification, suspense, and outcome resolution. Dev. Psychol. 20, 911–924. doi: 10.1037/0012-1649.20.5.911 Kawabata, H., and Zeki, S. (2004). Neural correlates of beauty. J. Neurophysiol. 91, 1699–1705. doi: 10.1152/jn.00696.2003 Koch, A., Krueger, H., and Unkelbach, C. (2016). A general valence asymmetry in similarity: good is more alike than bad. J. Exp. Psychol. Learn. Mem. Cogn. 42, 1171–1192. doi: 10.1037/xlm0000243 Koelsch, S., Jacobs, A. M., Menninghaus, W., Liebal, K., Klann-Delius, G., von Scheve, C., et al. (2015). The quartet theory of human emotions: an integrative and neurofunctional model. Phys. Life Rev. 13, 1–27. doi: 10.1016/j.plrev.2015.03.001 Kuchinke, L., Jacobs, A. M., Grubich, C., Võ, M. L. H., Conrad, M., and Herrmann, M. (2005). Incidental effects of emotional valence in single word processing: an fMRI study. Neuroimage 28, 1022–1032. doi: 10.1016/j.neuroimage.2005.06.050 Kuhlmann, M., Hofmann, M. J., Briesemeister, B. B., and Jacobs, A. M. (2016). Mixing positive and negative valence: affective-semantic integration of bivalent words. Sci. Rep. 6:30718. doi: 10.1038/srep30718 Kühn, S., and Gallinat, J. (2012). The neural correlates of subjective pleasantness. Neuroimage 61, 289–294. doi: 10.1016/j.neuroimage.2012.02.065 Kumaran, D., and McClelland, J. L. (2012). Generalization through the recurrent interaction of episodic memories: a model of the hippocampal system. Psychol. Rev. 119, 573–616. doi: 10.1037/a0028681 Larsen, S. E., and Seilman, U. (1988). Personal remindings while reading literature. Text 8, 411–429. doi: 10.1515/text.1.1988.8.4.411 Lebrecht, S., Bar, M., Barrett, L. F., and Tarr, M. J. (2012). Micro-valences: perceiving affective valence in everyday objects. Front. Psychol. 3:107. doi: 10.3389/fpsyg.2012.00107 Leder, H., Belke, B., Oeberst, A., and Augustin, M. D. (2004). A model of aesthetic appreciation and aesthetic judgments. Br. J. Psychol. 95, 489–508. doi: 10.1348/0007126042369811 Leder, H., and Nadal, M. (2014). Ten years of a model of aesthetic appreciation and aesthetic judgments: the aesthetic episode – developments and challenges in empirical aesthetics. Br. J. Psychol. 105, 443–464. doi: 10.1111/bjop.12084. Lehne, M., Engel, P., Rohrmeier, M., Menninghaus, W., Jacobs, A. M., and Koelsch, S. (2015). Reading a suspenseful literary text activates brain areas related to social cognition and predictive inference. PLoS ONE 10:e0124550. doi: 10.1371/journal.pone.0124550 Limbach, J. (2004). Das Schönste Deutsche Wort (The Most Beautiful German Word). Freiburg: Verlag Herder. Lüdtke, J., and Jacobs, A. M. (2015). The emotion potential of simple sentences: additive or interactive effects of nouns and adjectives? Front. Psychol. 6:1137. doi: 10.3389/fpsyg.2015.01137 Lüdtke, J., Meyer-Sickendiek, B., and Jacobs, A. M. (2014). Immersing in the stillness of an early morning: testing the mood empathy hypothesis in poems. Psychol. Aesthet. Creat. Arts 8, 363–377. doi: 10.1037/a0036826 Maddock, R. J., Garrett, A. S., and Buonocore, M. H. (2003). Posterior cingulate cortex activation by emotional words: fMRI evidence from a valence decision task. Hum. Brain Mapp. 18, 30–41. doi: 10.1002/hbm.10075 Mar, R. A. (2011). The neural bases of social cognition and story comprehension. Annu. Rev. Psychol. 63, 103–134. doi: 10.1146/annurev-psych-120709-145406

Marin, M. M. (2015). Crossing boundaries: toward a general model of 1768 neuroaesthetics. Front. Hum. Neurosci. 9:443. doi: 10.3389/fnhum.2015.00443 1769 Marin, M. M., Lampatz, I., Wandl, M., and Leder, H. (2016). Berlyne 1770 revisited: evidence for the multifaceted nature of hedonic tone in the appreciation of paintings and music. Front. Hum. Neurosci. 10:536. doi: 1771 1772 10.3389/fnhum.2016.00536 Martindale, C. (1975). Romantic Progression: The Psychology of Literary History. 1773 Washington, DC: Hemisphere. 1774 Martindale, C., and Dailey, A. (1995). I.A. Richards revisited: do people agree 1775 in their interpretations of literature? Poetics 23, 299–314. doi: 10.1016/03041776 422X(94)00025-2 Mason, R. A., and Just, M. A. (2009). The role of the theory-of-mind cortical 1777 network in the comprehension of narratives. Lang. Linguist. Compass 3, 1778 157–174. doi: 10.1111/j.1749-818X.2008.00122.x. 1779 McQuire, M., McCollum, L., and Chatterjee, A. (2016). Aptness and beauty in Q18 1780 metaphor. Lang. Cogn. doi: 10.1017/langcog.2016.13 1-16.DOI: 10.1017/langcog.20 Meyer-Sickendiek, B. (2011). Lyrisches Gespür: Vom Geheimen Sensorium 1781 Moderner Poesie [The Lyrical Sense of Feeling. About the Secret Sensorium of 1782 Modern Poetry]. Paderborn: Fink Publishing. doi: 10.1017/langcog.2016.13 1783 Miall, D. S. (1989). Beyond the schema given: affective comprehension of literary 1784 narratives. Cogn. Emot. 3, 55–78. doi: 10.1080/02699938908415236 1785 Miall, D. S., and Dissanayake, E. (2003). The poetics of babytalk. Hum. Nat. 14:337. 1786 doi: 10.1007/s12110-003-1010-4 Mieder, W. (2004). Proverbs: A Handbook. Westport: Greenwood Press. doi: 1787 10.1007/s12110-003-1010-4 1788 Morawetz, C., Bode, S., Baudewig, J., Jacobs, A. M., and Heekeren, H. R. (2016). 1789 Neural representation of emotion regulation goals. Hum. Brain Mapp. 37, 1790 600–620. doi: 10.1002/hbm.23053 Murphy, S. T., and Zajonc, R. B. (1993). Affect, cognition, and awareness: affective 1791 priming with optimal and suboptimal stimulus exposures. J. Pers. Soc. Psychol. 1792 64, 723–739. doi: 10.1037/0022-3514.64.5.723 1793 Nadal, M. (2013). “The experience of art: insights from neuroimaging,” in 1794 Progress in Brain Research, Vol. 204, eds S. Finger, D. W. Zaidel, F. Boller and J. Bogousslavsky (Amsterdam: Elsevier), 135–158. doi: 10.1037/0022- 1795 1796 3514.64.5.723 Nell, V. (1988). e psychology of reading for pleasure: needs and grati cations. Read. 1797 Res. Q. 23, 6–50. doi: 10.2307/747903 1798 Nicklas, P., and Jacobs, A. M. (2016). Rhetorics, neurocognitive poetics and the 1799 aesthetics of adaptation. Poetics Today doi: 10.2307/747903 Norris, C. J., Gollan, J., Berntson, G. G., and Cacioppo, J. T. (2010). The current 1800 status of research on the structure of affective space. Biol. Psychol. 84, 422–436. 1801 doi: 10.1016/j.biopsycho.2010.03.011 1802 Omigie, D. (2015). Music and literature: are there shared empathy and predictive 1803 mechanisms underlying their affective impact? Front. Psychol. 6:1250. doi: 1804 10.3389/fpsyg.2015.01250 Panksepp, J. (1998). Affective Neuroscience: The Foundations of Human 1805 and Animal Emotions. New York, NY: Oxford University Press. doi: 1806 10.3389/fpsyg.2015.01250 1807 Pehrs, C., Zaki, J., Schlochtermeier, L. H., Jacobs, A. M., Kuchinke, L., and 1808 Koelsch, S. (2015). The temporal pole top-down modulates the ventral visual stream during social cognition. Cereb. Cortex doi: 10.1093/cercor/bhv226 [Epub 1809 1810 ahead of print]. Pelowski, M., Markey, P. S., Lauring, J. O., and Leder, H. (2016). Visualizing the 1811 impact of art: an update and comparison of current psychological models of art 1812 experience. Front. Hum. Neurosci. 10:160. doi: 10.3389/fnhum.2016.00160 1813 Pleh, C. (2003). Narrativity in text construction and self construction. Neohelicon 1814 30, 187–205. doi: 10.1023/A:1024178827479 Pompe, A. (ed.) (2015). Kind und Gedicht (Child and Poem). Berlin: Rombach. doi: 1815 10.1023/A:1024178827479 1816 Ponz, A., Montant, M., Liegeois-Chauvel, C., Silva, C., Braun, M., Jacobs, A. M., 1817 et al. (2013). Emotion processing in words: a test of the neural re-use hypothesis 1818 using surface and intracranial EEG. Soc. Cogn. Affect. Neurosci. 9, 619–627. doi: 1819 10.1093/scan/nst034 Rangel, A., and Clithero, J. A. (2014). “The computation of stimulus values 1820 in simple choice,” in Neuroeconomics: Decision-Making and the Brain, 2nd 1821 Edn, P. Glimcher and E. Fehr (San Diego, CA: Academic Press). doi: 1822 10.1093/scan/nst034 1823

1767

1824


16


Jacobs et al.

1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858


Reber, R., Schwartz, N., and Winkielman, P. (2004). Processing fluency and aesthetic pleasure: is beauty in the perceiver’s processing experience? Pers. Soc. Psychol. Rev. 8, 364–382. doi: 10.1207/s15327957pspr0804_3 Richards, I. A. (1929). Practical Criticism: A Study of Literary Judgment. New York, NY: Harcourt Brace Jovanovich. doi: 10.1207/s15327957pspr0804_3 Schmidtke, D. S., Schröder, T., Jacobs, A. M., and Conrad, M. (2014). ANGST: affective norms for german sentiment terms, derived from the affective norms for english words. Behav. Res. Methods 46, 1108–1118. doi: 10.3758/s13428-0130426-y Schrott, R., and Jacobs, A. M. (2011). Gehirn und Gedicht: Wie wir unsere Wirklichkeiten Konstruieren (Brain and Poetry: How We Construct Our Realities). München: Hanser Publishers. doi: 10.3758/s13428-013-0426-y Silvia, P. J. (2009). Looking past pleasure: anger, confusion, disgust, pride, surprise, and other unusual aesthetic emotions. Psychol. Aesthet. Creat. Arts 3, 48–51. doi: 10.1037/a0014632 Simonton, D. K. (1990). Lexical choices and aesthetic success: a computer content analysis of 154 Shakespeare sonnets. Comput. Humanit. 24, 254–261. doi: 10.1037/a0014632 Simonton, D. K. (1989). Shakespeare’s sonnets: a case of and for single-case historiometry. J. Pers. 57, 695–721. doi: 10.1111/j.1467-6494.1989.tb00568.x Simonton, D. K. (1983). Dramatic greatness and content: a quantitative study of 81 Athenian and Shakespearean plays. Empir. Stud. Arts 1, 109–123. doi: 10.2190/0AGV-D8A9-HVDF-PL95 Stein, N. L. (1982). The definition of a story. J. Pragmat. 6, 487–507. doi: 10.1016/0378-2166(82)90022-4 Sternberg, M. (2003). Universals of narrative and their cognitivist fortunes (I). Poetics Today 24, 297–395. doi: 10.1215/03335372-24-3-517 Sylvester, T., Braun, M., Schmidtke, D., and Jacobs, A. M. (2016). The Berlin affective word list for children (kidBAWL): exploring processing of affective lexical semantics in the visual and auditory modalities. Front. Psychol. 7:969. doi: 10.3389/fpsyg.2016. 00969 Tsur, R. (1992). What Makes Sound Patterns Expressive? The Poetic Mode of Speech Perception. Durham, NC: Duke University Press. doi: 10.3389/fpsyg.2016.00969 Ullrich, S., Aryani, A., Kraxenberger, M., Jacobs, A. M., and Conrad, M. (2016). Textual features as basis of the general affective meaning: exploring the interplay of lexical affective features, dynamic inter-lexical shifts, and the basic affective tone in poetry. In revision. Unkelbach, C., Fiedler, K., Bayer, M., Stegmüller, M., and Danner, D. (2008). Why positive information is processed faster: the density hypothesis. J. Pers. Soc. Psychol. 95, 36–49. doi: 10.1037/0022-3514.95.1.36

Unkelbach, C., von Hippel, W., Forgas, J. P., Robinson, M. D., Shakarchi, R. J., and Hawkins, C. (2010). Good things come easy: subjective exposure frequency and the faster processing of positive information. Soc. Cogn. 28, 538–555. doi: 10.1521/soco.2010.28.4.538 Van den Hoven, E., Hartung, F., Burke, M., and Willems, R. (2016). Individual differences in sensitivity to style during literary reading: insights from eyetracking, Collabra. In revision. doi: 10.1521/soco.2010.28.4.538 van Peer, W. (1986). Stylistics and Psychology: Investigations of Foregrounding. London: Croom Helm. Võ, M. L. H., Conrad, M., Kuchinke, L., Urton, K., Hofmann, M. J., and Jacobs, A. M. (2009). The Berlin affective word list reloaded (BAWL-R). Behav. Res. Methods 41, 534–538. doi: 10.3758/BRM.41.2.534 Võ, M. L. H., Jacobs, A. M., and Conrad, M. (2006). Cross-validating the berlin affective word list. Behav. Res. Methods 38, 606–609. doi: 10.3758/BF Wainwright, J. (2011). Poetry: The Basics. London: Routledge. Westbury, C., Keith, J., Briesemeister, B. B., Hofmann, M. J., and Jacobs, A. M. (2014). Avoid violence, rioting, and outrage; approach celebration, delight, and strength: using large text corpora to compute valence, arousal, and the basic emotions. Q. J. Exp. Psychol. 68, 1599–1622, doi: 10.1080/17470218.2014.970204 Whissell, C. M. (1994). A computer program for the objective analysis of style and emotional connotations of prose: Hemingway, Galsworthy, and Faulkner Compared. Percept. Motor Skill 79, 815–824. doi: 10.2466/pms.1994.79.2.815 Wilensky, R. (1983). Story grammars versus story points. Behav. Brain Sci. 4, 579–623. doi: 10.1017/S0140525X00017520 Willems, R., and Jacobs, A. M. (2016). Caring about Dostoyevsky: the untapped potential of studying literature. Trends Cogn. Sci. 20, 243–245. doi: 10.1016/j.tics.2015.12.009 Wundt, W. (1896). Grundriss der Psychologie [Outline of Psychology]. Leipzig: Engelmann. Conflict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Copyright © 2016 Jacobs, Hofmann and Kinder. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

1859 1860 1861 1862 1863 1864 1865 1866

1882 1883 1884 1885 1886

1887press 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916

Kintsch, W. (2012). Musing about beauty. Cogn. Sci. 36, 635–654. doi: 10.1111/j. 1551-6709.2011.01229.x O’Sullivan, N., Davis, P., Billington, J., Gonzalez-Diaz, V., & Corcoran, R. (2015). “Shall I compare thee”: the neural basis of literary awareness, and its benefits to cognition, Cortex, 73, 144–157. doi: 10.1016/j.cortex.2015.08.014 Zeman, A., Milton, F., Smith, A., and Rylance, R. (2013). By heart: an fMRI study of brain activation by poetry and prose. J. Conscious. Stud. 20, 132–158.

1917 1918 1919 1920 1921 1922 1923

1867

1924

1868

1925

1869

1926

1870

1927

1871

1928

1872

1929

1873

1930

1874

1931

1875

1932

1876

1933

1877

1934

1878

1935

1879

1936

1880

1937

1881

1938


17


Q19