"STROOP" EFFECT: INPUT OR OUTPUT PHENOMENON?1 Two ...

1 downloads 0 Views 164KB Size Report
In two recent papers, this phenomenon—the. "Stroop" effect—has been given different interpre- tations. Keele (1972) has argued that selective attention must ...
Journal oj Experimental Psychology 1972, Vol. 95, No. 2, 458-459

"STROOP" EFFECT: INPUT OR OUTPUT PHENOMENON? 1 DOUGLAS L. HINTZMAN," FRANK A. CARRE, VERONICA L. ESKRIDGE, ANTHONY M. OWENS, STEPHANIE S. SHAFF, AND M. ELAINE SPARKS University of Oregon

Two explanations of the Stroop color-naming effect are: (a) that the form of the word and the color of the ink contact memory in parallel, causing covert response competition when they elicit different color names and (b) that color words distract from the ability to input or to encode ink color. In a mixed-list design, latencies of color-naming responses were taken for individual stimuli falling into four categories: (a) In S, the word and color name were the same; (6) in D, the word named a different color; (c) in N, the word was neutral with respect to color; (d) in C, the stimulus was a nonword. Compared to Cond. C, S responses showed facilitation, D showed interference, and N showed no effect. The facilitation in Cond. S is as predicted by the hypothesis that the Stroop effect is due to response competition rather than encoding interference. Keele (1972) in using a discrete-trial procedure and mixed-list design to avoid this problem. Unlike Keele's study, a vocal response was used. There were four conditions. In two of them the stimuli were color words that were either the same as or different from the required color-naming response. In a third the stimuli were neutral words, and in the control condition they were nonwords. Method.—Stimuli were produced by printing each word on a white card in one of four colors of ink: RED, GREEN, BLUE, or GOLD (yellow-orange). The letters were J in. square, and the lines approximately ^j in. wide. The stimuli fell into four categories: Cond. Same (S), in which the word named the color of ink in which it was printed; Cond. Different (D), in which the word named one of the three other colors; Cond. Neutral (N), which used noncolor words matched with color names in terms of first letter and word length, i.e., RAT, GLASS, BIRD, GLAD; and Cond. Control (C), which used anagrams of the color names with the first letters unchanged, i.e., RDE, GENER, BEUL, GDLO. Stimuli of Cond. S and D were generated by pairing each color name equally often with each ink color. Thus, Cond. D occurred three times as often as S, and color name was not correlated with color of ink. Altogether there were 128 experimental cards: 16 in Cond. S, 48 in D, 32 in N, and 32 in C. The stimulus sequence was divided into four blocks of 32 stimuli each (the order of presentation of these blocks was varied systematically across 5s). Each block, in turn, was divided into four subblocks. Each subblock consisted of 1 card in Cond. S, 3 in D, 2 in N, and 2 in C, presented in random order. Nineteen students in an undergraduate learning course at the University of Oregon served as 5s for extra course credit. The data of one S who made an unusually large number of errors and later confessed to being color-blind were discarded. Each S was seated before a tachistoscope and was told that words would be shown and that his task 1 This study was done as a Laboratory in Learning class was to say aloud, as quickly as possible, the color project. Thanks are due to Steven Keele for comments on the of ink in which each word was printed. Several manuscript. 2 Requests for reprints should be sent to Douglas L. Hintzexamples were shown during the instructions, to man, Department of Psychology, University of Oregon, Eugene, familiarize 5 with the procedure and with the four Oregon 97403. 458 If S is instructed to ignore the meaning of a word and to name as quickly as possible the color of ink in which the the word is printed, the amount of time taken to name the color depends on the word's meaning. If the word names a color other than the color of ink, the latency is greater than if the word is neutral with respect to color. In two recent papers, this phenomenon—the "Stroop" effect—has been given different interpretations. Keele (1972) has argued that selective attention must operate after, rather than before, memory retrieval. That is, the form and color of the stimulus are processed in parallel, and when these two sources of 'information elicit different color names, covert response competition results. This explanation localizes the Stroop effect at the output of the memory system. Hock and Egeth (1970), in contrast, localized the effect at input. Their explanation was that the nature of the task makes a color name particularly "attention-catching," so that it interferes with the ability to process and encode the color of ink. A way to discriminate between these two interpretations is to include a condition in which the color word and the color of ink match (e.g., BLUE in blue ink). If the Stroop effect is caused by response competition, then this condition should not produce interference. In fact, since both the form and the color of the stimulus lead to the same response, the color should be named faster than that of a neutral control word. On the other hand, if color words interfere with the ability to encode ink color, then the proposed condition should produce interference like that obtained when the color word does not match the color of ink. Traditionally, Stroop experiments have been done by manipulating conditions between lists (e.g., Pritchatt, 1968), a design that is subject to confounding by the use of different strategies in different conditions. The present experiment followed

459

SHORT REPORTS acceptable color names. Ten practice stimuli were then presented, followed by the 128 experimental stimuli. The stimulus duration was 1 sec., and the interstimulus interval approximately 4 sec. A vocal warning ("ready") was given just prior to the onset of each stimulus. Response latencies were measured using a .01-sec. stop clock triggered by a voice key. Latencies and errors were recorded by E. Results and discussion.—Only correct response latencies are presented. By conditions, the percentages of incorrect responses (including "false starts") were: S, 1.1; D, 13.2; N, 1.9; and C, 2.8. As would be expected, the classical Stroop condition produced the highest error rate. Every S's mean correct response latency was computed for each condition in each of the four blocks, and these means were submitted to a 4 X 4 repeated-measures analysis of variance. Stimulus conditions had a significant overall effect, F (3, 51) = 34.23, p < .001, and there was a slight decrease in latencies over blocks, F (3, 51) = 4.37, p < .01. The Conditions X Blocks interaction was marginally significant, reflecting a tendency for Cond. D to improve with practice more than the others, F (9, 153) = 2.41, p < .05. Overall mean latencies by conditions (in msec.) were: S, 653; D, 799; N, 694; C, 696. Planned comparisons showed that Cond. S was significantly faster than Cond. C, t (17) = 3.08, p < .01, and Cond. D significantly slower than C, t (17) = 6.15, p < .001. Conditions N and C did not differ significantly. This lack of a difference between Cond. N and C is consistent with Keele's (1972) results, suggesting that the effect of neutral words on color

naming found by Klein (1964) is eliminated when the discrete-trial procedure is used. The facilitation found in Cond. S tends to confirm Keele's (1972) claim that selective attention operates after memory retrieval. If the classic Stroop condition were due primarily to interference with the ability to encode ink color, then it should make little difference whether or not a color word matches the color of ink; interference should be produced in either case. The present results indicate that a compatible color word enhances color-naming performance not only in comparison with the classic Stroop condition, but also in comparison with the neutral control. This is what would be expected if the Stroop effect were due to covert response competition. Of course, one cannot be certain that color words produced no encoding interference in the present experiment. However, if such interference was present, its effect must have been relatively small. The present findings argue that the Stroop effect is primarily an output, rather than an input phenomenon. REFERENCES HOCK, H. S., & EGETH, H. Verbal interference with encoding in a perceptual classification task. Journal of Experimental Psychology, 1970, 83, 299-303. KEELE, S. W. Attention demands of memory retrieval. Journal of Experimental Psychology, 1972, 93, 245-248. KLEIN, G. S. Semantic power measured through the interference of words with color-naming. American Journal of Psychology, 1964, 77, 576-588. PRITCHATT, D. An investigation of some of the underlying associative verbal processes of the Stroop color effect. Quarterly Journal of Experimental Psychology, 1968, 20, 351-359. (Received September 23, 1971)