The Semantic Simon Effect - SAGE Journals

3 downloads 0 Views 88KB Size Report
Three experiments are reported in which a semantic variant of the Simon ..... Proctor & T.G. Reeve (Eds.), Stimulus±response compatibility: An integrated ...
T HE QUARTERLY JOURNAL OF EXPERIMENTAL PS YCHOLOGY, 1998, 51A (3), 683± 688

T h e Sem an tic Sim on Effect Jan De Houwer Univ ersity of Leuv en, Leuv en, Belgium

T hree experiments are reported in which a semantic variant of the Simon paradigm was used. In Experiment 1, participants saw Dutch and English words that corresponded to names of animals (e.g. DOG ) or occupations (e.g. T EACHER). Participants were instructed to respond by saying ANIMAL or OCCUPATION, depending on whether the presented word was a Dutch or English word (i.e. relevant stimulus feature) but irrespective of whether the word was the name of an animal or an occupation (i.e. irrelevant stimulus feature). Results showed that responses were facilitated when the correct response corresponded to the name of the semantic category of the presented word (e.g. saying ``ANIMAL ’’ to DOG ) compared to when it was the name of a different semantic category (e.g. saying ``OCCUPAT ION’ ’ to DOG ), even though the semantic category of the presented word was irrelevant and had to be ignored. Category membership also in¯ uenced response times when letter case (upper- or lower-case: Experiment 2) and grammatical category (noun or adjective: Experiment 3) had to be determined in order to select a category label as a response. T he semantic Simon effect offers a new tool that can be used to study automatic semantic processing.

In a typical Simon experiment, participants are instructed to make a spatial response on the basis of a non-spatial stimulus feature. Importantly, the relevant stimulus is presented lateralized. Despite the fact that the spatial location of the relevant stimulus is irrelevant, responses are facilitated when the spatial position of the stimulus corresponds to the spatial features of the response that has to be made (Simon, 1990). For instance, participants might be asked to press the left key of a button-press device upon presentation of a red light and the right key upon presentation of a green light, irrespective of whether the light is presented on the left or the right side of a visual display. In this case, responses to the red light will be facilitated when it is presented on the left side of the display compared to when it is presented on the right side. Responses to the green light will be facilitated when presented on the right side (Craft & Simon, 1970). T hree elements play an important role within the Simon paradigm: a relevant stimulus feature that determines what the correct response should be (e.g. the colour of the light),

Requests for reprints should be sent to Jan De Houwer, Department of Psycholog y, University of Leuven, T iensestraat 102, B-3000 Leuven, Belgium. E-mail: Jan.DeHouwer@ps y.kuleuven.ac.be Jan De Houwer is a post- doctoral researcher for the Fund for Scienti® c Research (Flanders-Belgium). He is now at Department of Psychology, University of S outhampton , High® eld, Southampton SO17 1BJ, U.K. I thank Paul Eelen, Frank Baeyens, Geert Crombez, Dirk Hermans, Herman Buelens, Bernhard Hommel, and an anonymous reviewer for their helpful comments on an earlier draft of the paper.

q

1998 T he Experimental Psycholog y S ociety

684

DE HOUWER

an irrelevant stimulus feature that has to be ignored (e.g. the spatial location of the light), and a relevant response feature (e.g. the spatial properties of the keypress responses). T he irrelevant stimulus feature is related to the relevant response feature, whereas the relevant stimulus feature is not related to either the relevant response feature or to the irrelevant stimulus feature. In the case of spatial Simon effects, the irrelevant stimulus feature (e.g. left or right presentation of a stimulus) and the relevant response feature (e.g. left or right keypress) are related because they both incorporate spatial information. T here is no such similarity or overlap between the relevant stimulus feature (e.g. colour of the light) and the relevant response feature or the irrelevant stimulus feature (Kornblum, 1992). In the present experiments, I used a new variant of the S imon paradigm in which the irrelevant stimulus feature and the relevant response feature were similar with regard to their semantic meaning. Words were presented that belonged to one of two semantic categories and also differed with regard to a second, relevant stimulus feature. Participants were instructed to respond by saying the name of one of two semantic categories and could select the correct response on the basis of the relevant stimulus feature. In Experiment 1, the language of the words (Dutch or English) was relevant, letter case (upper- or lower-case) was relevant in Experiment 2, and in Experiment 3, the grammatical category of the words (noun or adjective) was relevant. T he purpose of the present experiments was to explore whether reliable semantic Simon effects can be observed. A semantic Simon effect would be evidenced by faster or more accurate responses on trials where the semantic category of the presented word (irrelevant stimulus feature) and the meaning of the correct response (relevant response feature) correspond (congruent: e.g. saying ``ANIMAL’ ’ to the word HORSE) compared to trials where there is no such correspondence (incongruent: e.g. saying ``OCCUPAT ION ’ ’ to the word HORSE).

Method Participants In Experiment 1, 16 ® rst-year students (13 females, 3 males) participated, in Experiment 2, 32 ® rst-year students (25 females, 7 males), and in Experiment 3, 16 second-year students (12 females, 4 males). All participants were native Dutch speakers and participated for partial ful® lment of course requirements.

Materials In Experiment 1, 24 names of animals and 24 names of occupations were used as stimuli. Four animal and occupation names were presented during practice trials; the other names were used on experimental trials. All words were written in upper-case letters. Half of the animal and occupation names were Dutch words (e.g. HOND [ DOG ], TIMMERMAN [ CARPENTER]), half were English words (e.g. HORSE , T EACHER). All Dutch and English words referred to a different animal or occupation. In Experiment 2, 24 animal names and 24 names of occupations were selected, 4 of each category as practice stimuli and the remainder as experimental stimuli. All names were Dutch words. Half of the names of each category were written in upper-case letters, half in lower-case letters. In Experiment 3, 30 nouns and 30 adjectives were presented in upper-case letters. Ten nouns and adjectives were presented during a practice block; the remaining words functioned as experimental stimuli. Half of

SEMANTIC SIMON EFFECT

685

the practice and experimental nouns referred to animals (e.g. VLINDER [ BUT TERFLY]), half referred to persons (e.g. VRIEND [ FRIEND ]). Adjectives were not speci® cally related to either animals or persons. All stimuli were presented on a 70-Hz SVGA screen connected to a 486 PC. Words were presented in white letters on a black background. Each letter was 7 mm high and 5 mm wide. Presentations were controlled by a Turbo Pascal 5.0 program that operated in graphics mode. Participants were seated in front of the computer screen at a distance of approximately 40 cm. Verbal responses were registered using a voice key which generated a signal that stopped a highly accurate Turbo Pascal timer (Bovens & Brysbaert, 1990).

Procedure In Experiment 1, half of the participants were instructed to say the Dutch word DIER [ ANIMAL ] upon presentation of a Dutch word and the Dutch word BEROEP [ OCCUPAT ION ] upon presentation of an English word, whereas the other participants were to say DIER to English words and BEROEP to Dutch words. In Experiment 2, half were asked to say DIER when the presented word was written in uppercase letters and BEROEP when it was presented in lower-case letters; the reverse was true for the other participants. In Experiment 3, where names of animals and persons were presented, half of the participants were to say DIER to nouns and PERSOON [ PERSON] to adjectives, whereas half were to say PERSOON to nouns and DIER to adjectives. Participants were informed that words would correspond to names of animals and occupations (Experiments 1 and 2) or that some words would correspond to names of animals and persons (Experiment 3), but they were told that the semantic category of the words was unimportant and should be ignored. In Experiments 1 and 2, the eight practice words were presented twice in a random order during a practice block. In Experiment 3, all 20 practice stimuli were presented once in a random order. In all experiments, the practice block was followed by two blocks of 40 experimental trials. In each experimental block, each of the 40 experimental stimuli was presented once in a random order with the restriction that the language (Experiment 1), letter case (Experiment 2), or grammatical category (Experiment 3) of the presented word could not be the same on four or more consecutive trials. Each practice and experimental trial consisted of the following sequence of events: a warning tone (1000 Hz, 200 msec), ® xation cross (500 msec), blank screen (500 msec), and a word, which remained on the screen until the voice key registered a response. If no response was registered, the word disappeared after 3000 msec. All visual stimuli were presented in the middle of the computer screen. After each trial, the experimenter entered a code that corresponded to the response that was given, except on trials where the voice key failed to register the response correctly, in which event a different code was entered. T he next trial was initiated 1500 msec after the experimenter entered a code. At the end of Experiment 1, participants were asked to give the Dutch translation of the presented English words.

Results Experiment 1 Trials were coded as congruent (when the correct response word corresponded to the name of the semantic category of the presented word) or incongruent (when the correct response was the name of the other category). T he percentage of errors and the mean reaction time (RT ) on congruent and incongruent trials was determined for each participant (across stimuli) and for each stimulus (across participants) in the following way. Only experimental trials were taken into account. Trials on which a voice key failure

686

DE HOUWER

occurred or an English word that was unknown to the participant was presented, were excluded. RTs on trials where an incorrect response was given were also discarded. T he signi® cance level was set at p < .05. Voice key failures occurred on 2.81% of all trials. Incorrect responses were given on 4.02% of the remaining trials. On 2.66% of all trials, an English word was presented that was unknown to a participant. Means can be found in Table 1. An ANOVA performed on data collapsed across stimuli showed that RTs were shorter on congruent than on incongruent trials, F(1, 15) = 21.09, MSE = 1512.60 , and revealed a tendency for fewer errors on congruent than on incongruent trials, F(1, 15) = 3.03, MSE = 16.92, p = .10. Item analyses con® rmed these results: F(1, 39) = 8.07, MSE = 3940.60 , for RTs; F(1, 39) = 6.79, MSE = 22.60, for errors. When language (Dutch or English) was introduced as a variable in the item analyses, results showed that the effect of congruence was not mediated by the language of the words (Fs < 1.15).

Experiment 2 Means (Table 1) were calculated in the same way as in Experiment 1. Voice key failures occurred on 5.28% of all trials, and incorrect responses were given on 1.78% of all remaining trials. ANOVA showed that RTs were shorter on congruent than on incongruent trials, both when participants, F(1, 31) = 8.05, MSE = 1629.59 , and stimuli, F(1, 39) = 5.76, MSE = 1763.83 , were treated as the random factor. Analyses of the error data did not reveal any effects (Fs < 1).

Experiment 3 Because only nouns were related to the responses, only trials on which a noun was presented were taken into account. Apart from this difference, means (Table 1) were calculated in the same way as in the previous experiments. Voice key failures occurred on 2.42% of all trials, and incorrect responses were given on 2.56% of the remaining trials. When participants were treated as the random factor, ANOVA showed that responses were faster on congruent than on incongruent trials, F(1, 15) = 6.42, MSE = 7228.92 . Item analyses of RT data also revealed a signi® cant

TABLE 1 Mean RT and Percentage Error on Congruent and Incongruen t Trials Congruent RT Experiment 1 2 3

Incongruent Error Rate

RT

Error Rate

Msec

SD

%

SD

Msec

SD

%

SD

665 584 678

80 103 98

2.77 1.77 1.49

3.94 2.52 2.04

710 605 730

92 105 135

5.30 1.80 2.12

5.23 2.14 3.10

SEMANTIC SIMON EFFECT

687

main effect of congruence, F(1, 19) = 7.79, MS E = 2632.03 . Analyses of the error data did not reveal any effects (Fs < 1).

Discussion T he core of the S imon paradigm is that participants have to respond to a relevant stimulus feature while ignoring an irrelevant stimulus feature that is related to a relevant response feature (Kornblum, 1992). Whereas in most studies using the Simon paradigm, the irrelevant stimulus feature and the relevant response feature were spatially related, in the present experiments both features were semantically related. T he reported results showed that responses were in¯ uenced by the match between the meaning of the correct response and the semantic category of the words, even though the latter was irrelevant and had to be ignored. T he fact that a signi® cant semantic Simon effect was observed in three consecutive experiments, each involving a different relevant stimulus feature, attests to its reliablity and generality. T he semantic Simon effect provides an additional tool to study automatic semantic processing. Moreover, it has some advantages compared to related paradigms, such as the semantic priming paradigm (e.g. Neely, 1991; see Boucart & Humphreys, 1992, for a related paradigm) and the S troop paradigm (Stroop, 1935), which until now have been used to study automatic semantic processing. For instance, in semantic priming and Stroop studies, targets and distractors that are semantically related are most often also related on other non-semantic dimensions (e.g. associatively or perceptually). Due to such confounding, it is often dif® cult to establish whether priming and Stroop effects actually re¯ ect automatic semantic processing of distractors rather than non-semantic processing (e.g. LaHeij, 1988; Shelton & Martin, 1992; Williams, 1996). Using the semantic Simon effect, one can integrate the relevant (target ) and irrelevant (distractor) stimulus into one stimulus and because the relevant and irrelevant stimulus features are always unrelated, confounding as described above cannot occur. T he semantic S imon effect thus allows one to minimize the risk of confounding in a straightforward manner while offering a maximum of ¯ exibility. Moreover, whereas priming and Stroop studies that controlled for confounding of semantic and non-semantic variables often revealed only small effects, semantic Simon effects seem to be strong and reliable.

REFERENCES Boucart, M., & Humphreys, G.W. (1992). Global shape cannot be attended without object identi® cation. Journal of Experimental Psychology: Human Perception and Performance, 18, 785± 806. Bovens, N., & Brysbaert, M. (1990). IBM PC/ XT / AT and PS/ 2 Turbo Pascal timing with extended resolution. Behavior Research Methods, Instruments, and Computers, 22, 332± 334. Craft, J.L., & Simon, J.R. (1970). Processing symbolic information from a visual display: Interference from an irrelevant directional cue. Journal of Experimental Psychology, 83, 415± 420. Kornblum, S. (1992). Dimensional overlap and dimensional relevance in stimulus-response and stimulusstimulus compatibility. In G.E. Stelmach & J. Requin (Eds.), Tutorials in motor behavior, Vol. 2 (pp. 743± 777). Amsterdam: North Holland.

688

DE HOUWER

LaHeij, W. (1988). Components of Stroop-like interference in picture naming. Memory and Cognition, 16, 400± 410. Neely, J.H. (1991). Semantic priming effects in visual word recognition: A selective review of current ® ndings and theories. In D. Besner & G.W. Humphreys (Eds.), Basic processes in reading: Visual word recognition (pp. 264± 336). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. S helton, J.R., & Martin, R.C. (1992). How semantic is semantic priming? Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 1191± 1210. Simon, J.R. (1990). T he effects of an irrelevant directional cue on human information processing. In R.W. Proctor & T.G. Reeve (Eds.), Stimulus± response compatibility : An integrated perspectiv e (pp. 31± 86). Amsterdam: North Holland. Stroop, J.R. (1935). S tudies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643± 662. Williams, J.N. (1996). Is automatic priming semantic? European Journal of Cognitive Psychology, 8, 113± 161. Original manuscript receiv ed 30 June 1997 Accepted revision received 13 February 1998