Culture and visual perception

9 downloads 650178 Views 126KB Size Report
of a 12-inch laptop computer (Macintosh iBook). .... .17. Relevant means were shown in Figure 2. Americans were more accurate than Japanese in the two part ...
Culture and Visual Perception

Japanese Psychological Research (in press)

Culture and Visual Perception: Does Perceptual Inference Depend on Culture?1

Keiko Ishii

Takafumi Tsukasaki

Hokkaido University

and

Shinobu Kitayama University of Michigan

Correspondence regarding this manuscript should be sent to: Keiko Ishii Center for Experimental Research in Social Sciences Hokkaido University N10W7, Kita-ku, Sapporo 060-0810 [email protected] Phone: 81-11-706-3056

1

Culture and Visual Perception

Abstract

Some perceptual cues carry information about the overall pattern of an object (holistic cues), whereas others carry information about the distinct parts of an object (part cues). Drawing on recent work on culture and cognition, the authors predicted that people with European-American cultural backgrounds would be more capable than those with Asian backgrounds of utilizing part cues in perceptual inference. No such cross-cultural difference was expected for the ability to utilize holistic cues. In two studies, participants were presented with either one of the two types of cues and asked to infer the identity of the original objects. As predicted, in the part cue condition European-Americans performed better than Japanese (Study 1) and Asian-Americans (Study 2) did. Also as predicted, there was no cultural difference in the holistic cue condition. The results were interpreted in reference to other related studies documenting reliable cross-cultural differences in cognition. (146 words)

Key words: culture, perception, and inference

2

Culture and Visual Perception

Imagine that you saw something in front of you. Further imagine that the object moved so quickly that you could not recognize what it was. Naturally, you engage in effortful perceptual inferences. Such inferences are quite ubiquitous in daily life. Not surprisingly, they have also been used as an important research tool in investigating psychological mechanisms underlying perceptual information processing (Erdelyi, 1974; Niedenthal & Kitayama, 1994; Neisser, 1967). The present work follows this literature and extends it cross-culturally. We predicted that there would be a cross-cultural difference in the ability to make accurate perceptual inferences. This difference, however, would be moderated by the type of perceptual cues that are available. Culture and Perception A number of culturally oriented psychologists have recently revived the thesis of “New Look” in the 1950’s (Bruner, 1957) and begun intensive research programs on how cultural lay theories and practices might influence relatively low-level perception and attention (Kitayama, Duffy, Kawamura, & Larsen, 2003; Nisbett, Peng, Choi, & Norenzayan, 2001). The researchers have hypothesized that in Western cultures people tend to focus attention toward a distinct object and its attributes in lieu of the broader context in which the object is embedded. In contrast, in East Asian cultures people tend to attend more holistically to both an object and its context. In agreement with an earlier proposal by Sloman (1996), focused attention – which is more characteristic of Westerners – may lend itself to analytic reasoning, whereas holistic attention – which is more characteristic of Asians – may be conducive to associative retrieval of information. We suggest that systematic, linear inferences require focused attention even when the inferences are essentially perceptual in nature. Initial evidence for this analysis comes from a study by Masuda and Nisbett (2001). These researchers found that when accounting for an underwater scene, Americans were more attentive to focal objects, referring to attributes and actions of these objects. In contrast, Japanese were much more attuned to the context or the whole field and reported the focal 3

Culture and Visual Perception

objects only secondarily. Kitayama et al. (2003) extended the Masuda and Nisbett research to perception of geometric figures and showed that Americans are more capable of attending to a focal object while ignoring its surrounding, whereas Japanese are more capable of attention to the whole scene. Perceptual Inference The present paper reports two experiments designed to examine cultural variations in perceptual inference. By extending the literature on cultural differences in attention, we tested whether the accuracy in perceptual inference might vary across cultures depending on the type of perceptual cues provided. When an object such as basketball, clock, or tiger is presented under suboptimal viewing conditions, two conceptually distinct types of cues can exist simultaneously. Some cues (called holistic cues) suggest an overall shape or pattern associated with the object at hand whereas some other cues (called part cues) suggest specific parts that stand out. For example, imagine that something moved very quickly at the periphery of your sight. You might recognize the overall shape or pattern of the moving object as beast-like even though you might not be able to tell exactly what that is. You might also recognize some very specific stimulus elements, such as yellow and black stripes. In combination, you might correctly guess that the moving object must be a tiger. The two types of cues may require somewhat different kinds of inferences if they are to successfully lead to correct identification of the impinging object. Imagine, first, that part cues are available. In such circumstances, people will have to identify what these cues are and then to reconstruct a whole object of which these cues are constitutive parts. Such perceptual inferences would require an ability to attend to parts and combine them to yield a meaningful pattern, thereby reconstructing the whole object. We hypothesized that focused attention, which is characteristic of Euro-Americans, would be required both to pay close attention to parts and to draw systematic inferences from them. Our first prediction was that perceptual inference would be more accurate for Euro-Americans than for Asians when it is 4

Culture and Visual Perception

based on part cues. Now imagine that only a holistic cue such as overall shape or an extremely degraded image of the object at hand is available. Under such conditions, people will have to fill in details within the overall blurry image to arrive at the correct identity of the object at hand. This type of inference would require an ability to holistically attend to the entire pattern. At the same time, however, like any other inferences, if they are to be systematic, these inferences would also require analytic processing, with attention focused on a sequence of reasoning that is made from the cue to the identity of the object. Accordingly, holistic attention may result in no overall advantage as compared to more focused attention. Our second prediction, then, was that perceptual inference would be equally accurate for both Asians and Euro-Americans when it is based on holistic cues.

Study 1. Perceptual Identification by Americans and Japanese Method Participants and procedure.

Thirty Japanese undergraduates (12 females and 18

males) at Kyoto University and 23 American undergraduates (9 females and 14 males) at Kansai Foreign Language University (Osaka, Japan) participated in the study.2 The American participants were spending several months in Japan for training in the Japanese language. Participants received 500 yen for their participation. They were tested individually in a classroom of their respective universities. Upon arrival, participants were seated in front of a 12-inch laptop computer (Macintosh iBook). A chin-rest was placed 45 cm from the computer. Each of the original objects was fitted to a rectangular area of 21.6 cm in width and 14.5 cm in height on the computer monitor, resulting in a horizontal visual angle of approximately 27º. The rooms were lighted with fluorescent lamps. Participants were told that they would be shown a number of common objects under very restricted viewing conditions. They were told that their task was to identify what these objects were. All instructions were given in their native languages. 5

Culture and Visual Perception

Each trial started with the presentation of a trial number on the computer monitor. When participants pressed the space bar, a fixation point (+) appeared at the center of the screen. When participants pressed the space bar again, a stimulus picture was presented for 3 seconds. They were to examine the stimulus carefully, infer the identity of the original object, and report the name on a sheet of paper. Participants were encouraged to write down their best guesses although they were allowed to skip the trial if it was impossible for them to even guess. After the 3 s exposure of the stimulus, the next trial number automatically appeared on the screen. This was repeated 90 times. These 90 trials were preceded by six practice trials. Materials.

We first prepared color photographs of 96 objects, which were

presented to separate groups of 41 Japanese and 41 American undergraduate participants. These participants judged how frequently they saw each object in daily life (1= “not at all”, 5 = “very frequently”). We chose 30 objects (e.g., apple, tiger, and pinecone) that were fairly familiar (M > 2.07) and, moreover, whose familiarity scores did not show any cross-cultural difference. For each object, we created three color stimuli to be used. A single-part stimulus was a single square-shaped part of the original photograph. Care was taken to use a part that was diagnostic of the original object. A multiple-part stimulus was composed of several smaller square shaped parts of the original photograph. The parts were chosen so that they were diagnostic of the original object. We included the two part cue conditions for exploratory purposes. Because in both cases one must identify distinct elements and draw inferences from them, we predicted no difference between the two conditions. A holistic stimulus is a blurred version of the original picture (Figure 1). The mosaic module of Adobe Photoshop was used to create the holistic stimuli. A series of informal pilot tests were conducted to adjust both the diagnosticity of the part(s) chosen and the blurredness of the Gestalt so that the ease of identifying the original objects was roughly equal for all the 30 objects. This procedure yielded 90 stimuli (30 objects x 3 stimulus types). All the 90 stimuli were presented within a counter-balancing scheme. First, the 30 objects were arbitrarily divided into three sets (A, B, and C). The total of 90 trials was 6

Culture and Visual Perception

divided into three blocks of 30 trials each. Within each block, the three sets of stimuli were assigned to the three cue conditions (single part, multiple part, and holistic) so that over the three blocks any given set of stimuli were used once in each of the three cue conditions. Within each block the 30 stimuli were randomly ordered. The resulting set of 90 stimuli was presented in either the order created this way or its reverse. Results For each participant we computed the proportion of objects that were correctly identified for each of the three stimulus conditions of each of the three blocks.3 We then performed an ANOVA on the percent correct measure with three between-subjects variables (culture [Japanese and Americans], gender [females and males], block [first, second, and third]) and one within-subjects variable (cue type [single part, multiple part, and holistic]). Overall, a significant cue type main effect showed that accuracy was highly for the part cue conditions than for the holistic cue condition, F(2, 82) = 68.37, p < .0001, ηp2 = .63. This effect suggests that our effort to equate the difficulty or ease of identification across the three cue types was not completely successful. In general, the part cue stimuli were more effective as a cue for identification than the holistic cue stimuli. It is important to note, however, that when accuracy was computed for each of the 90 stimuli, there was considerable overlap in accuracy across the three cue conditions. Importantly, Americans were more accurate than Japanese, F(1, 41) = 5.26, p < .05,

ηp2 = .11. Moreover, as predicted, the interaction between culture and cue type proved significant, F(2, 82) = 8.22, p < .001, ηp2 = .17. Relevant means were shown in Figure 2. Americans were more accurate than Japanese in the two part cue conditions, Ms = 0.60 vs. 0.50, t(123) = 2.59, p < .01, Cohen’s d = .75 and Ms = 0.63 vs. 0.50, t(123) = 3.49, p < .01, Cohen’s d = .92, for the single part and the multiple part conditions, respectively. In contrast, this cross-cultural difference was non-significantly reversed in the holistic cue condition, Ms = 0.39 vs. 0.38, t(123) = 0.13. The same pattern was observed equally across the three blocks. The interaction involving culture, cue type, and block was negligible. 7

Culture and Visual Perception

Study 2. Extension with Between-Subjects Design Study 1 provides initial evidence that the accuracy of perceptual inference depends both on culture and cue type. When part cues were available, Americans were more accurate than Japanese. Importantly, we found no cross-cultural difference in the holistic cue condition. Nevertheless, it seemed important to replicate it in a more tightly controlled experimental condition. First, participants might not have expended sufficient effort to perceptual identification when the cues presented are very ambiguous. Because it was easier on average to identify objects from part cues than to do so from holistic cues, participants might not have expended enough effort when holistic cues were presented and the task seemed especially difficult. This might seem especially likely in Study 1 where the cues of all the three types were mixed within a single session. In Study 2, therefore, we manipulated cue type as a between-subjects variable. Second, one important shortcoming of Study 1 stemmed from the fact that Japanese and American participants were tested in two different testing locations. Although lighting was seemingly equivalent in the two locations and, moreover, we used the same computer for stimulus presentation, there might be subtler differences in the conditions of the rooms that confounded the cultural differences or the absence thereof. Study 2 therefore tested two ethnic groups (Euro vs. Asian Americans) in a single US location. This arrangement enabled us to address whether Asian Americans would show the same pattern as did Japanese in Study 1. Because Study 1 found no difference between the two part cue conditions, Study 2 examined only one of them -- the multiple part condition. Method Eighty-four undergraduates (50 females and 34 males) at University of Michigan participated in the study. Thirty-one out of the 84 participants were of Asian origin.4 Their mean length of stay in the US was 12.2 years (SD = 8.1). The other participants were European Americans. They received $8 for their participation. They were tested individually. Upon arrival, participants were seated in front of a 12-inch laptop computer (Macintosh 8

Culture and Visual Perception

iBook). They were then instructed in English that their task was to identify what the original pictures were based on some cues provided. The procedure was the same as that in Study 1 with one exception. In Study 2 participants were presented with either the multiple part stimuli or the holistic stimuli in all the three blocks of 30 trials in each. The experimental trials were preceded by 4 practice trials. Stimulus pictures were presented with the same order used in Study 1 so that each picture was presented three times. Any given stimulus was never presented in succession. Results We computed the proportion of correct responses for each participant separately for each of the three blocks.5 The resulting proportions were then submitted to an ANOVA with culture (Asian and European), block (first, second, and third) and cue type (multiple part and holistic). As in Study 1 we found a significant main effect of cue type, with accuracy higher in the part cue condition than in the holistic condition, F(1, 80) = 16.97, p < .0001, ηp2 = .18. Also as in Study 1, the culture and cue type interaction proved significant, F(1, 80) = 4.98, p < .05, ηp2 = .06. In the part cue condition, European Americans were more accurate than Asians, Ms = 0.56 vs. 0.43, t(80) = 2.81 p < .01, Cohen’s d = .80. However, in the holistic cue condition there was no cultural difference, Ms = 0.37 vs. 0.35, for Asian and European Americans respectively, t(80) = 0.32, ns. Finally, we found no correlations between accuracy and length of stay in the Asian American participants (r = .24 and .10, ns., for the multi part and the holistic pictures, respectively).

General Discussion In agreement with the hypothesis that Americans are focused in attention and analytic in reasoning, they were reliably more capable of drawing accurate inferences based on part cues than Japanese did. We also hypothesized that Asians are holistic in attention, meaning that they are more capable of attending to the whole and, yet, relatively disadvantaged in engaging in analytic inferences. In agreement with this analysis, in both 9

Culture and Visual Perception

experiments reported here, we found no cross-cultural difference in the holistic cue condition. It is worthy of note that the pattern shown by Japanese in Study 1 could be generalized to Asian Americans in Study 2. Clearly, culture cannot be neatly mapped onto geographic locations in which people live. It is possible that Asian Americans are socialized in such a way that they acquire cognitive tools of “seeing the forest” whereas European Americans appear to acquire the cognitive tools of “scrutinizing the trees”. These habits of attention are likely to reflect different socialization goals of culture. These socialization goals are known to be quite variable across different cultures (Keller, 2003). They may also vary across different individuals within any given culture. Sub-group differences within a given broad cultural region such as Asia and the U.S. must be carefully assessed in future work (see e.g., Kitayama, Ishii, Imada, Takemura, & Ramaswamy, 2006, for an initial effort in this direction). We should hasten to add that the cultural difference observed here is a matter of degree. Thus, both Asians and European Americans are entirely capable of utilizing both types of perceptual cues. Yet, it is also important that the efficiency of processing depends on both the cultural backgrounds of participants and cue types. This, perhaps, means that the basic architecture of perceptual processes is no different across cultures, but parameters are set in such a way that persistent cross-cultural variations accrue over a number of different tasks and domains. It appears that the range of attention is set to be much wider in Asian cultures than in Western cultures. Yet, the more precise nature of such parameter settings has yet to be examined in terms of eye movement (Chua, Boland, & Nisbett, 2005) and the neural processing units that are recruited (Park & Gutchess, 2006). Furthermore, it will be important to extend the current work to other related attentional and perceptual phenomena such as visual search and visual illusions. Through such investigations, it may be possible to further explore the role of socio-cultural experiences in shaping and even forming different strategies or styles of processing even at relatively low levels of perception and attention. 10

Culture and Visual Perception

Reference Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64, 123-152. Chua, H. F., Boland, J. E., & Nisbett, R. E. (2005). Cultural variation in eye movements during scene perception. Proceedings of the National Academy of Sciences of the United States of America, 102, 12629-12633. Erdelyi, M. H. (1974). A new look at the new look: Perceptual defense and vigilance. Psychological Review, 81, 1-25. Keller, H. (2003). Socialization for competence: Cultural models of infancy. Child Development, 46, 288-311. Kitayama, S., Duffy, S., Kawamura, T., & Larsen, J. (2003). Perceiving an object and its context in different cultures: A cultural look at the new look. Psychological Science, 14, 201-206. Kitayama, S., Ishii, K., Imada, T., Takemura, K., & Ramaswamy, J. (2006). Voluntary settlement and the spirit of independence: Evidence from Japan’s “Northern Frontier”. Journal of Personality and Social Psychology, 91, 369-384. Masuda, T., & Nisbett, R. E. (2001). Attending holistically versus analytically: Comparing the context sensitivity of Japanese and Americans. Journal of Personality and Social Psychology, 81, 922-934. Neisser, U. (1967). Cognitive Psychology. New York: Appleton-Century-Crofts. Niedenthal, P., & Kitayama, S. (1994). The Heart's Eye: Emotional Influences in Perception and Attention. New York: Academic Press. Nisbett, R. E., Peng, K., Choi, I., & Norenzayan, A. (2001). Culture and systems of thought: Holistic vs. analytic cognition. Psychological Review, 108, 291-310. Park, D., & Gutchess, A. (2006). The cognitive neuroscience of aging and culture. Current Directions in Psychological Science, 15, 105-108. Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119, 3-22. 11

Culture and Visual Perception

Footnotes 1

Study 1 is based on an undergraduate thesis submitted by Takafumi Tsukasaki under the

guidance of Shinobu Kitayama to Kyoto University. The research was supported by a Japan Society for the Promotion of Science Fellowship to Keiko Ishii. 2

Nobody reported that she or he had color-blindness in either Study 1 or Study 2.

3

A Japanese-English bilingual coder coded all the responses, and a Japanese coder and an

American coder coded the Japanese and American responses respectively. Responses were considered correct if they referred to the identity of the object at issue or its semantic equivalents. Highly general answers (e.g., “map” for “globe”) were coded as incorrect. Agreement was quite high for both American and Japanese data (Cohen’s kappa’s = 0.96 and 0.92 respectively). 4

The origins were as follows: Chinese (14), Indian (6), Korean (3), Vietnamese (1), and

Philippine (1). The other 6 people were unknown. 5

Two coders coded responses to each trial in the same way as in Study 1. Agreement was

quite high (Cohen’s kappa = 0.96).

12

Culture and Visual Perception

Figure Captions

Figure 1. An example (lobster) of stimulus pictures. For each object, we created three types of stimuli (single part, multi part, and holistic pictures).

Figure 2. Accuracy in the judgment for single part, multi part, and holistic pictures in Japanese and Americans (Study 1). Americans were more accurate than Japanese in the two part cue conditions, whereas this pattern vanished in the holistic cue condition.

13

Culture and Visual Perception

Original object

Single part

Multiple part

Figure 1

14

Holistic

Culture and Visual Perception

Accuracy 0.7

Accuracy

0.6

0.5

Japanese Japanese Americans Americans 0.4

0.3

0.2 Single part Part Single

Multiple Part Multiple part

Figure 2

15

Gestalt Holistic