Dissociative Affective and Associative Priming Effects

Journal of Experimental Psychology: Learning, Memory, and Cognition 2000, \W. 26, No. 2 , 4 5 6 - * 9

Copyright 2000 by the American Psychological Association, Inc. 0278-7393/00/$5.00 DOI: 10.1037//0278-7393.26.2.456

Dissociative Affective and Associative Priming Effects in the Lexical Decision Task: Yes Versus No Responses to Word Targets Reveal Evaluative Judgment Tendencies Dirk Wentura University of Mtinster The affective priming effect (AP; i.e., shorter evaluative or lexical decision latencies for affectively congruent prime-target pairs) has often been interpreted as evidence for spreading activation from the prime to affectively congruent targets. The present study emphasizes the view that in the lexical decision task, the prime-target configuration is implicitly evaluated as a question of the form "Is (prime) (target)?" (e.g., "Is death wise?") so that there is a tendency to affirm in cases of congruency and to negate in cases of incongruency. Therefore, after establishing the AP with the lexical decision task in Experiment 1, in Experiment 2 the assignment of yes responses to words and nonwords was varied. For the word = yes condition, the AP emerged, whereas the data pattern was reversed for the word = no condition. In Experiment 3, a comparable pattern of results was not found for symmetrical or backward associativelyrelatedprime-target pairs.

Recently, there has been a growing interest in the processing of affective-evaluative information within the cognitive system. Fazio and colleagues (e.g., Fazio, Sanbonmatsu, Powell, & Kardes, 1986) were instrumental in stimulating research on what can be called affective priming (see also Bargh, Chaiken, Govender, & Pratto, 1992; Bargh, Chaiken, Raymond, & Hymes, 1996; Greenwald, Klinger, & Liu, 1989; Hermans, 1996; Hermans, De Houwer, & Eelen, 1994; Klauer, RoBnagel, & Musch, 1997; Rothermund & Wentura, 1998; Wentura, 1994; for an overview see Klauer, 1998). On the basis of the tradition of social psychological research on attitudes, Fazio et al. (1986) sought to confirm the attitude accessibility hypothesis (Fazio, 1986), according to which the mere presentation of highly valenced objects or words automatically increases the accessibility of the corresponding affective evaluation. To find support for their hypothesis, they used an experimental paradigm based on the semantic priming paradigm (e.g., Neely, 1977; Neely, 1991). Fazio et al. presented their participants positively and negatively valenced adjectives (e.g., repulsive, appealing) on a monitor and asked them to classify these target words as positive or negative as quickly as possible (evaluation task). Shortly before each target (300 ms), they presented a positively or negatively valenced prime word (e.g., death, gift). The authors corroborated their hypothesis by showing that the decision latency for the target words was shorter

I thank Alexandra Grainer, Thomas Graw, Harald Hormann, Marion Kalmbach, and Marita Schmitt (Experiment 1), and Uta Jaschinski (Experiments 2 and 3) for their help in conducting the experiments. I thank David Bunnedi and Corrine Voils for their help in improving the English of this article. Correspondence concerning this article should be addressed to Dirk Wentura, Psychologisches Institut IV, University of Minister, Fliednerstrasse 21, D-48149 Miinster, Germany. Electronic mail may be sent to [email protected]. 456

when the prime and target were congruent in terms of valence than when they were not Theoretically, there are two foci of attention in affective priming research. The first concerns the question of whether the valence of a stimulus will be automatically activated on mere presentation of that stimulus. Although there has been considerable debate on this topic (cf. Bargh et al., 1992; Chaiken & Bargh, 1993; Fazio, 1993; Fazio et al., 1986; Greenwald et al., 1989; Hermans, 1996; Klauer, 1998), it seems fair to conclude that this assumption—i.e., that evaluations are automatically activated—is no longer disputed (see, e.g., recent studies by Draine & Greenwald, 1998; Greenwald, Draine, & Abrams, 1996). The second focus of affective priming research concerns the mediating mechanisms by which the target evaluation is influenced by the prime presentation. This point was somewhat neglected in earlier studies but has recently gained importance (Bargh et al., 1996; Hermans et al., 1994; Klauer et al., 1997; Wentura, 1994, 1997, 1999) and lies at the very heart of the current research. Methodologically, the domain of affective priming research can be structured along three dimensions. The first one relates to the task that is used for assessing the affective congruency effect. Besides the evaluation task which is used in only this research area, three other tasks that are well-known in semantic priming research have been used. Participants have been required to make a word-nonword decision regarding the target (the lexical decision task), to say the target aloud (the pronunciation task), or to name the color of the target (the Stroop-priming task). The second dimension refers to the materials used as primes and targets. Coming from the tradition of social psychological research on attitudes, most studies on affective priming used nouns (referring to attitude objects) as primes and adjectives as targets. However, some studies used adjective-adjective pairs (e.g., Klauer et al., 1997), noun-noun pairs (e.g., Bargh

DISSOCIATIVE AFFECTIVE AND ASSOCIATIVE PRIMING

et al., 1996, Experiment 3), mixtures of nouns and adjectives (e.g., Greenwald et al., 1989), a picture-picture variant (e.g., Hermans et al., 1994), or odors as primes (e.g., Hermans, Baeyens, & Eelen, 1998), emphasizing the supposed generalizability of the mechanisms underlying affective priming effects. A trurd dimension is the stimulus onset asynchrony (SOA) of prime and target. In order to limit the complexity of this exposition, further discussion here is restricted to short SOA conditions (300 ms or less). It is known from research on semantic priming that effects emerging under SOA conditions longer than 500 ms may be affected by slow-acting intentional processes (i.e., that given the prime word, participants anticipate the target word; see Neely, 1977; for an overview see Neely, 1991). These processes are of only minor interest in the present context. With these factors in mind, the results of affective priming studies can be summarized as follows: With short primetarget SOAs of 300 ms or less, affective priming effects occur in most studies using an evaluation task. This applies especially to studies using noun-adjective pairings (Bargh et al., 1992; Chaiken & Bargh, 1993; Fazio et al., 1986; Wentura, 1994) as well as to the picture-picture experiment of Hermans et al. (1994). Using only adjectives or a mixture of nouns and adjectives seems to result in effects for only very short SOAs (i.e., up to 100 ms; Greenwald, Klinger, & Schuh, 1995; Klauer et al., 1997; but see Greenwald et al., 1989). For the lexical decision and pronunciation tasks, mixed results have been obtained. Hill and Kemp-Wheeler (1989; Kemp-Wheeler & Hill, 1992) provided some evidence for affective priming in the lexical decision task with noun-noun pairs. Unfortunately, those authors compared only negative to neutral primes with respect to their effect on negative targets. Using adjective-adjective pairs, Klauer, Rofinagel, and Musch (1995) obtained no effects. The pronunciation task was used successfully by Bargh et al. (1996) and Hermans et al. (1994), whereas Hermans (1996, Experiment 8), De Houwer, Hermans, and Eelen (1998), and Klauer and Musch (1997) failed to obtain affective congruence effects. In the Stroop-priming task (modeled after Warren, 1972, 1974), no evidence was found for affective congruence effects (Hermans, Van den Broeck, & Eelen, 1998; Rothermund & Wentura, 1998). The Spreading Activation Model Drawing on the similarity of affective priming to semantic priming, the concept of spreading activation in a semantic network has been used as an explanation for affective congruence effects. According to this theory (e.g., Anderson & Pirolli, 1984; Collins & Loftus, 1975), priming effects are explained through the mechanism of spreading activation operating in a network of nodes (symbolizing concepts) and links (symbolizing associations or semantic relations). Thus, presentation of the word butter, for example, activates linked nodes (e.g., bread), and this activation makes the information more accessible for further processing (e.g., reading bread). Similarly, Fazio et al. (1986) argued that Presentation of an attitude object would automatically activate any strong association to that object. Such activation is

457

assumed to spread along the paths of the memory network, including any evaluative associations. Consequently, the activation levels of associated evaluations are temporarily increased. If a target word that corresponds in valence to one of these previously activated evaluations is subsequently presented for judgment, then less additional activation is required for the activation level of the target word [italics added] to reach threshold and, consequently, for a judgment to be made. (P-231) Garnering empirical support for this assumption is important for understanding the mental representation of affective information. For example, an often-cited representational approach in the field of cognition and emotion is the associative network theory of emotions by Bower (1981). Essentially, Bower added emotion nodes to the semantic network. The mechanism of spreading activation can account for the triggering of cognitive, physiological, and experiential aspects of emotion (for revised formulations, see Bower, 1987, 1991; Bower & Cohen, 1982). Thus, Hill and Kemp-Wheeler (1989; Kemp-Wheeler & Hill, 1992), arguing in the same spirit as Fazio et al. (1986), classified affective priming effects as variants of semantic priming phenomena. Kitayama and Howard (1994) viewed semantic priming in the tradition of Bower to be one of two dominant themes in theories of affective information processing. De Houwer and Hermans (1994; see also Hermans et al., 1994) proposed a related approach to the storage of affective information in semantic memory. According to this theory, nodes representing positive or negative concepts are connected to global positivity and negativity nodes, respectively. Thus, presentation of a positive or negative word activates the corresponding node; activation spreads via the global positivity or negativity node to all nodes with the same evaluation. Subsequently, those concepts are more accessible, and thus a corresponding word is processed faster. This "localist" model of representation might be rejected on a priori reasons: Usually the amount of spreading activation is assumed to be limited (e.g., Anderson, 1974, 1983). Thus it seems rather implausible that spreading activation from one positive concept will activate all other positive concepts to a detectable degree (see also Hermans, De Houwer, & Eelen, 1996). However, in distributed models of representation (e.g., McClelland, Rumelhart, & PDP Research Group, 1986; Rumelhart, McClelland, & PDP Research Group, 1986), the basic idea of Fazio et al. (1986) might be preserved without implausible assumptions. Take, for example, Masson's (1995) model, which proposes that semantic priming effects are due to the overlap of the representational patterns of prime and target. Updating the activation pattern on target presentation will be shortened if a related prime has instantiated a somewhat similar pattern just before (see also McRae, de Sa, & Seidenberg, 1997). Supposing that a subset of nodes in the network corresponds to the valence of a concept would allow the prediction of an affective congruence effect. According to both variants of the spreading activation model, processing of the target is necessary only for assessing its heightened accessibility. The underlying process, spreading activation, has taken place whether there is a

458

WENTURA

target or not In principle, therefore, spreading activation can account for affective priming effects in all tasks, because to assess the altered memory state, it is necessary merely to apply a task that demands semantic processing of the target. In this respect, the spreading activation model does not distinguish between processing affective and nonaffective stimuli. "Affective" means nothing more than belonging to the semantic category of positive or negative information. However, drawing an analogy between affective priming and semantic priming overlooks the fact that to date unequivocal evidence for affective congruence effects has been gathered with the evaluation task only. This task, however, differs in an important respect from the tasks typically used in semantic priming paradigms. In the lexical decision task, for example, the prime-target relationship (i.e., semantically-associatively related vs. not related) is not confounded with the prime-response relationship: Both types of primes (i.e., related vs. unrelated ones) are words and will therefore not differ in their response compatibility. In the evaluation task, however, the prime-target relationship (i.e., affectively congruent vs. incongment) is perfectly confounded with the prime-response relationship: Whenever a prime is congruent with the target, its valence matches the response that is needed. However, whenever it is incongruent with the target, there is also a mismatch to the required response. If, for example, a positive target has to be classified as "positive!" a congruent prime is also positive, thereby predicting the correct response, whereas an incongruent one is negative, thereby potentially biasing the wrong answer. Because of these factors, the affective congmency effect could be due to response facilitation or interference rather than to spreading activation. In other words, in the evaluation task, presentation of a prime might trigger a response process due to its affective value. Therefore, when prime and target are affectively congruent, facilitation of the correct response may occur. An even more likely event is interference in cases of incongruency (cf. Bargh et al., 1996; Klauer et al., 1997; Wentura, 1994). In this regard, the evaluation task has much more in common with the Stroop-interference paradigm (Stroop, 1935; for a review see MacLeod, 1991) than with the semantic priming paradigm. In fact, derivations following from this analogy can be confirmed (see Klauer et al., 1997; Musch & Klauer, 1997; Wentura, 1997, 1999). Klauer et al. (1997) showed that affective priming effects are determined by the proportion of congruent to incongruent trials. At short SOAs, this occurs in Stroop tasks but not in associative priming tasks. Wentura (1999) studied the influence of congruence or incongruence on a first trial on the response in a subsequent trial. If mere is a response path conflict in incongruent trials, this should result in a negative priming effect like those that are known from Stroop-like paradigms (for reviews, see Fox, 1995; May, Kane, & Hasher, 1995; Neill, Valdes, & Terry, 1995). In accordance with this assumption, an affective incongruence of prime and target on a first trial resulted in a slowing of the response on the next trial if that target was affectively congruent to the prime of the former trial. In sum, there are clear reasons to believe that affective

congruency effects in the evaluation task are the result of Stroop-like processes. But, as mentioned above, there is some evidence of affective priming effects in the lexical decision task. Response path facilitation-interference mechanisms cannot be easily used to explain the priming effects in the lexical decision task, because congruent and incongruent primes seemingly do not differ in their relevance for the task of classifying target stimuli as words or nonwords. Judgmental Tendency Mechanisms An alternative explanation to the spreading activation account can be made for affective congruence effects in the lexical decision task, at least for noun-adjective pairings. Klauer and Stem (1992; Klauer, 1991) proposed a model of affective-cognitive links in judgment processes as an explicit alternative to the spreading activation account. According to this model, when judging statements of the form "(noun) is (adjective)" (e.g., "Einstein is fond of animals"), a three-component process will take place. First, the affective components of both the noun and the adjective are activated. Second, the compatibility of the affective components are compared. Both processes are supposed to be automatic. The result of the second process is an a priori hypothesis about the correct answer. Therefore a judgmental tendency to affirm (in cases of affective congruency) or to reject (in cases of incongruency) the relation in question results. Third, in a controlled process, relevant information is recalled to form an adequate answer on the basis of the a priori hypothesis and the available information. An experimental procedure to test the hypothesis that affective components are automatically compared was constructed by Klauer and Stern (1992). In that procedure, participants were required to classify pairs of stimuli (the name of a politician paired with a trait word, two names, or two traits; pairs were presented simultaneously) according to some formal characteristic ("Does this pair consist of a person and a trait?"), requiring a yes-no response (actually, the authors used the labels "correct" and "false"). Politicians as well as traits were evaluated as positive or negative so that affective congruency could be varied. In accordance with the model, response latencies for person-trait pairs were facilitated in cases of affective congruency and inhibited in cases of incongruency. But, by reversing the question so that person-trait pairs required a no response ("Does mis pair consist of either two persons or two traits?")* the pattern of latencies was reversed (Klauer & Stern, 1992). This second-order interaction was critical in supporting the theory. It should be obvious that the theory can be applied to the lexical decision paradigm. The affective congruency effect in the lexical decision task may be a mere by-product of the judgmental tendencies caused by the compound of prime and target, given the assumption that the lexical decision task is implicitly coded as requiring yes (word) and no (nonword) responses (see West & Stanovich, 1982, regarding this assumption).


Overview Until now, the evidence for affective priming effects in the lexical decision task has been somewhat sparse. Therefore, in Experiment 1, prime and target valence were varied in the lexical decision task to establish this variant of affective priming. Experiment 2 was the crucial test for the judgmental tendency model. In a between-subjects factorial design, the word targets in the lexical decision task were assigned a yes or no response. The spreading activation model would predict no second-order interaction (i.e., the affective congruency effect should not depend on response assignment), whereas the affective congruency effect should hold only under the yes condition but be reversed under the no condition according to the judgmental tendency model. To rule out alternative hypotheses, in Experiment 3 the yes-no assignment was varied with associatively related primetarget pairs. The hypothesis is that yes-no assignment makes no difference in associative priming effects.

Experiment 1 Experiment 1 tested for an affective congruency effect in the lexical decision task. If the often-observed affective congruency effect in the evaluation task is not solely due to response path facilitation or interference, an effect of prime-target congruence should also emerge in the lexical decision paradigm.

Method Participants. Fifty-five students (37 women; 18 men) at a German university participated in the experiment in partial fulfillment of course requirements. The median age was 23.0 years (range: 19 to 28 years); 3 participants were replaced because they made more than 20% errors (for word or nonword responses). Design. Central to the hypothesis of affective congruency is the factorial combination of target valence (positive vs. negative) and prime valence (positive vs. negative). A neutral condition consisting of neutrally valenced words served as a third priming condition (see Materials). For exploratory reasons, I established a fourth prime condition that consisted of pseudonouns (see Materials). As adding this condition establishes an additional 2 (word targets vs. pseudoword targets) X 2 (word primes vs. pseudoword primes) design, it is possible to look for response facilitation or interference due to the match or mismatch of prime category and target response. The four priming conditions were crossed with four groups of participants (13 to 14 per group) and four sets of stimuli (see Materials) in a Latin-square design to ensure that each prime and target was presented only once to each participant. Each type of prime (positive, negative, neutral, or nonword) was followed by a word or nonword an equal number of times. Materials. Four sets of 10 positive and 10 negative German adjectives each were selected as targets on the basis of their pleasantness values (absolute values above 49 on a scale ranging from - 1 0 0 to +100) according to a norm list of 908 adjectives (Hager, Mecklenbriiuker, MOller, & Westermann, 1985; Mffller & Hager, 1991). Each set included 20 pronounceable nonwords that had endings like those of German adjectives (e.g., pargnant and kriftagnich). Priming stimuli were 40 positive nouns, 40 negative nouns, 40 neutral nouns, and 40 pseudonouns (e.g., SlRTAL, VELMEN). The positive and negative stimuli had already been

459

used in experiments by Wentura (1999). Neutral prime words were selected and classified a priori, but they were also validated by participants in a rating task carried out after the lexical decision task. Positive, negative, and neutral prime words had mean ratings of 2.14 (SD = 0.40), -1.95 (SD = 0.61) and 0.31 (SD = 0.36), respectively, on a scale of - 3 (very negative) to + 3 (very positive). Because affectively congruent word pairs seem to be more frequently semantically related than incongruent ones, the random assignment of primes to targets may result in a certain rate of semantic related pairs per participant, with the consequence of a moderate semantic priming effect. Thus, the assignment of positive, negative, neutral, and nonword primes to target stimuli was fixed to guarantee semantic or associative unrelatedness. Target categories and sets, as well as prime categories and sets, were balanced according to stimulus length. Because of coding errors, one positive target from each set had to be removed, resulting in 9 instead of 10 values to be aggregated in the positive target condition. Procedure. Participants were seated in front of an IBMcompatible microcomputer and on the CRT screen received instructions regarding the task. They were instructed to classify the target stimuli as legitimate words or nonwords. Both speed and accuracy were emphasized. Participants were told that every trial would start with one preparatory stimulus (a word or nonword) followed by a target stimulus that must be categorized. The prime (in uppercase letters) was presented for 200 ms, followed by a 100-ms interval before the onset of the target. Therefore, the stimulus onset asynchrony (SOA) was 300 ms. Choice of these parameters followed Fazio et al. (1986). The target disappeared as soon as participants pressed a key. The interval between the response and the beginning of the next trial was 2,500 ms. Word and nonword decisions were assigned to the right and left index finger, respectively. (The "a" and " # " keys on a German standard keyboard were used.) Latency of response was recorded to the nearest millisecond (see Heathcote, 1988). After 20 practice trials, 40 positive, 40 negative, and 80 nonword targets were presented, with the order of presentation randomized for each participant. Following the lexical decision task, participants were asked to rate the affective connotation of the prime words on a scale from —3 (very negative) to + 3 (very positive; sec Materials). Results Mean reaction times (RTs) were derived from correct responses only. The average error rate across participants was 2.68% and 3.89% for words and nonwords, respectively. RTs that were three interquartile ranges above the third quartile with respect to the individual distribution (see Tukey, 1977) or were above 1,500 ms were discarded as well (1.99% and 4.09% of all word and nonword decisions, respectively). For the sake of simplicity and clarity, I computed mean RTs with reference to the prime-target relation. In other words, responses to positive-positive and negative-negative prime-target pairs were aggregated (congruent priming) as well as negative-positive and positive-negative primetarget pairs (incongruent priming). For the two neutral conditions, data were collapsed across positive and negative targets. (Target valence did not significantly affect the results reported below.) All analyses were performed using both participant (referred to by tu Fx) and item (t%, F2) means; means were adjusted for the effects of (he material lists that were used for counterbalancing (see Pollatsek & Well,

460

WENTURA

1995). Unless otherwise noted, all effects referred to as statistically significant throughout the text are associated with/? values less than .05, two-tailed. Results for the conditions of interest are shown in Table 1. Hie pattern of means is congruent with the affective priming effect found by Fazio et al. (1986). An analysis of variance (ANOVA) revealed a significant overall effect of primetarget relation, F,(3,162) = 5.87, MSE = 801; F 2 (3,225) = 5.88, MSE = 1,074. A priori planned contrasts were conducted. Of main interest is the comparison of the congruent and incongruent conditions; this contrast (d = 0.32) proved to be significant, ft(54) = 2.36, t2(75) = 2.17. The mean RT for the neutral condition (with words as primes) fell between those for the congruent and incongruent conditions; t tests, however, did not reveal significant differences between the neutral and congruent or incongruent conditions (all |f|s < 1.35). The 16-ms difference between RTs to word targets in the two neutral conditions was significant, *i(54) = 2.69, *2(75) = 2.96. Because the corresponding 12-ms difference for nonword targets was in the same direction and of comparable magnitude—though significant only for the within-subjects analysis, ^(54) = 2.07; f2(79) = 1.82, p < .10—mis effect is probably due to prime processing being more difficult for the nonwords than for neutral words rather than being due to any sort of response bias. As this effect is not theoretically important in the current context, it is not discussed further. The error data showed a similar but weaker pattern. Neither the analysis of variance (both Fs < 1.20) nor the comparison of most interest (congruent vs. incongruent priming) revealed any significant effects (both | r|s < 1.27).

the spreading activation account and the judgmental tendency hypothesis. Assigning the word targets in the lexical decision task to either yes or no responses should make no difference according to a spreading activation account, whereas the affective congrueney effect should hold under only the yes condition but be reversed under the no condition, according to the judgmental tendency hypothesis. At this point of discussion it is necessary to prevent a misunderstanding. Spreading activation and judgmental tendencies do not preclude each other. Facilitation of access and facilitation-inhibition of response operate at different stages of information processing. Thus, there is no reason to believe that the magnitude of facilitory effects due to spreading activation—if there are any—depends on judgmental tendencies. Therefore, both processes might have contributed to the congrueney effect that was observed in Experiment 1. Thus, the pattern of results observed in Experiment 2 could reveal a mixture of both processes. Specifically, if judgmental tendencies were making no contribution to affective priming and spreading activation was the sole source of affective priming in Experiment 1, a spreading activation account predicts that a congrueney effect of exactly the same magnitude should occur in the word — no condition of Experiment 2. However, if judgmental tendencies and spreading activation both contributed to the affective priming effect observed in Experiment 1, then affective priming in the word = no condition should be less positive than in the word = yes condition because the judgmental tendency process would now favor the incongruent priming condition in the word = no condition and offset facilitation due to spreading activation. More specifically, if the facilitation due to spreading activation and the judgmental tendency effects in Experiment 1 were of identical magnitudes, they would cancel each other out in the word = no condition, yielding an affective priming effect of 0. If the facilitation due to spreading activation was greater in absolute value than the judgmental tendency effect, then one should still see a positive affective priming effect in the word = no condition but one smaller than that observed in the word ~ yes condition. If facilitation due to spreading activation was smaller in absolute magnitude than the judgmental tendency effect, then one should now see an inhibitory affective priming effect in the word = no condition that has an absolute value smaller than the facilitatory priming effect seen in the word = yes condition. Finally, if spreading

Discussion The hypothesis of affective priming in the lexical decision task was confirmed: After presentation of negative primes (compared with positive primes), the responses to negative targets were facilitated; this pattern was reversed for positive targets. The means for the neutral prime condition were in between, with a slight resemblance to the positive prime condition, which, incidentally, corresponds to the distance of the mean rating from zero for the neutral primes (see Materials). After showing a significant affective congrueney effect in the lexical decision task, we can now turn to testing between

Table 1 Mean Response Times (in ms) and Error Rates (in %) for Affectively Polarized Targets as a Function of Prime-Target Relation (Affectively Congruent vs. Incongruent vs. Neutral; Experiment 1) AP8

Priming Variable

Congruent

Incongruent

Neutral (words)

Neutral (nonwords)

M

SE

Reaction times (ms) Error rates {%)

639 2.3

650 2.9

645 2.1

661 2.9

U 0.6

5 0.6

•Affective priming: Mean of incongruent minus congruent prime condition; standard errors.


activation was making no contribution to the affective priming effect in Experiment 1 and the judgmental tendency was the sole source of the affective priming effect, then there should be an inhibitory effect in the word = no condition of exactly the same absolute magnitude as the facilitator,' effect in the word = yes condition. Experiment 2 The spreading activation account and the judgmental tendency hypothesis make the same predictions in cases where an affirmative response mode is used; they diverge, however, when a negative response mode is used. Assigning word stimuli to either yes or no responses in the lexical decision task is the critical test. In a first attempt, Wentura (1998) found that the affective congmency effect in the word = yes condition switches into an incongruency effect in the word = no condition. There are, however, some caveats concerning this study. First of all, the effect was rather small, statistically significant only in a one-tailed / test, though the sample size was large (N = 91). The second caveat refers to the error data with an indication of a speed-accuracy trade-off for the word = yes condition (i.e., there was a negative priming effect for error rates). Finally, one may speculate whether the result in the word — no condition might be caused by a two-fold violation of expectancies: Participants not only had to respond with no to words, they also had to press the right-hand key in those cases. If most participants would naturally assign affirmative answers to the right-hand key, this might distort the results. Therefore, the assignment of response mode to the right- and left-hand keys should be varied independently of the word = yes versus word — no factor.

Method Participants. Forty-seven native speakers of German (mainly students; 43 women, 4 men) participated. Psychology students took part in partial fulfillment of course requirements; small awards of money were raffled off to the remaining participants. The median age was 21.0 years (range: 17 to 43 years). Four persons were replaced because of their self-reported failure to comply with instructions (see below). The data of one other person were excluded because of an extremely high error rate (i.e., above 20%). Design. Two between-subjects factors were added to the design of Experiment 1. Most important, response assignment was varied. In one condition (n = 25), participants were asked to respond with yes to target words (word = yes condition). In the other condition (n = 22), they were asked to respond no to target words (word = no condition). In addition, hand assignment was varied orthogonal to response assignment. Responses to words had to be given either with the right-hand key or the left-hand key. To enhance power for the contrast of interest, I discarded the nonword and neutral prime conditions so that mean RTs for congruent and incongruent priming could be based on more trials. Materials. Materials were essentially the same as in Experiment 1. To enhance statistical power, material sets were increased in size to 24 positive and 24 negative targets per set, totaling 96 word targets and 96 nonword targets. Thus, mean RTs for congruent and incongruent priming were now based on 48 trials each (for the analysis with participants as the random variable).

461

Following Experiment 1 and Wentura (1998), I replaced target adjectives with the highest error rates. For each participant, each prime word was presented twice—one time preceding a word and one time preceding a nonword target. Primes were now presented in standard German form (i.e., first character in uppercase, the remaining characters in lowercase); I replaced prime words with the German letter 6 because of their irregular graphical representation on the CRT screen. A three-step procedure ensured the associative and semantic unrelatedness of primes and targets. In the first step, participants (N = 15) wrote down the first adjective that came to mind with regard to the prime word (with the restriction of not turning the stem of the noun into an adjective). For affectively incongruent prime-target pairs, no match occurred; for affectively congruent pairs, one match out of a total of 1,440 (96 words X 15 participants) responses occurred. In a second step, I asked a sample of participants (N — 15) to write down the first noun that came to mind with regard to the target adjective (with the restriction of not turning the stem of the adjective into a noun). Neither for affectively incongruent prime-target pairs nor for affectively congruent pairs did a match occur. Li a third step, I used the following procedure to test for the semantic (un-)relatedness of congruent and incongruent prime-target pairs. Participants (N = 4) were instructed to rate noun-adjective pairs for semantic relatedness. A high degree of semantic relationship was explicated as one of the following: (a) that the ascription of the adjective to the noun object is very likely (e.g., "A JOKE is [usually] FUNNY"), (b) that there is a high degree of semantic overlap (e.g., "MADNESS means to be CRAZY"), or (c) that there is a contextual proximity (e.g., "To play the VIOLIN, one has to be MUSICAL"). On each trial, one noun and three adjectives appeared on the computer monitor. Besides the affective congruent and incongruent target adjectives, the adjective most frequently associated with the noun (in Step 1) was presented to make the task somewhat meaningful to the participants. For each noun-adjective pair, a rating was given on a scale from 0 (no semantic relationship) to 8 (strong semantic relationship). Whereas associated adjectives showed a strong semantic relationship to their respective nouns (Af = 7.2l; SD = 0.99), this relationship was missing for the incongruent (Af = 0.58; SD = 0.94) and congruent (Af= 1.28; SD = 1.20) adjectives.1 Procedure. The procedure was essentially the same as in Experiment 1. There were three minor changes: (a) Responses were now given recorded by response boxes connected via an I/O port to the computers, thereby reducing potential error variance due to using the standard keyboard; (b) a hardware timer (with millisecond precision) was used; and (c) participants were tested in small groups but in individual cubicles. Instructions (given on the CRT screen) now emphasized the respective yes category ("You have to respond with YES to words [or to nonwords], otherwise with NO"). To counteract the natural tendency to view words as "figure" and nonwords as "ground," nonwords were now given a Gestalt-like character by naming them "SIWOBs" (a German acronym for "senseless wordlike strings") in the word ~ no condition and by giving some examples. The ease of detecting SIWOBs was emphasized. To practice this assignment, participants completed 10 practice 1 Although the difference between incongruent and congruent adjectives is negligible compared with the difference between associated and incongruent-congruent adjectives, it was significant for the within-items analysis, *,(3) = 2.26, ns; t2(95) = 4.36. This might be due to the inability of participants to refrain from using affective congruence as a sign of semantic relatedness.

462

WENTURA

trials in which they simply categorized the target; that is, no prune was presented. Feedback was given after each trial (i.e., for the word - no condition "Right! It was [not] a SIWOB. You responded correctly with YES [NO]!" or "Wrong! It was [not] a SIWOB. You should have responded with YES [NO]!"; for the word = yes condition "Right! It was [not] a WORD . . , " or "Wrong! It was [not] a WORD . . . " ) . If the response was slower than 1,000 ms, the following message was added: "Your reaction time was (RT) ms. Please try to respond more quickly!" Next, the presentation of the prime stimuli was explained. Participants were told that each trial would begin with a first stimulus, followed by a target stimulus that they must categorize. This procedure was also practiced over 10 trials. For the practice trials, participants were encouraged to accompany key presses by saying "yes!" or "no!" to themselves. The main phase began with 20 practice trials, as in Experiment 1. The order of presentation was randomized for each participant. As mentioned in the Materials section, each prime was presented twice, one time preceding a word and one time preceding a nonword target. For each participant and for each prime, the first (second) presentation was nearly equally often in the congruent and incongruent conditions. After the priming task, participants rated how well they felt they followed the instructions on a 7-point scale ranging from 0 (/ could not adhere to this instruction, I always regarded the responses as "word" and "nonword" responses.) to 6 (/ had no problems following the instructions. 1 always regarded my responses as "yes1*and "no"responses.). Data fiom participants who responded with 0(N = 4) were discarded (see Participants).

Results

incongruent) analysis of variance. The result of central interest is the significant interaction between yes-no assignment and prime-target relation, corresponding to a switch of an affective congruency effect in the word = yes condition to an incongruency effect in the word — no condition, F,(l, 45) = 4.09, MSE = 260; F 2 (l, 95) = 4.93, MSE = 1,199. The main effect of yes-no assignment was also significant for the within-items analysis, Fi(l, 45) = 3.10, MSE = 13,598, p .10. Responses to the targets used in the symmetrical condition were faster than those to the backward-related targets; all others Fs < 1. The error data revealed essentially the same pattern as the RTs. There was a significant main effect of prime-target relation, F ^ l , 46) = 7.13, MSE = 18.31; F 2 (l, 94) = 6.90, MSE = 37.82. The priming effect was 1.65% (d = 0.39). There was also a main effect of yes-no assignment in the within-items analysis, F 2 (l, 94) = 5.02, MSE = 41.61; but F v (l, 46) = 2.45, MSE = 42.70, p > .10; all other Fs < 1.50. Overall, a robust priming effect emerged for associatively related prime-target pairs. As predicted by the judgmental tendency model, this effect was not moderated by response assignment. General Discussion In summary, the main line of results and arguments is as follows: First, affective congruency effects (Fazio et al., 1986) are sometimes explained by some sort of spreading activation account—either framed in the traditional seman-

465


Table 3 Mean Response Times (in ms) and Error Rates (in %) for Neutral Targets as a Function of Prime-Target Relation (Associatively Related vs. Unrelated), Response Assignment, and Direction of Association (Experiment 3) AP8

Priming Variable Symmetrical priming Reaction times (ms) Word = yes Word = no Overall Error rates (%) Word = yes Word = no Overall Backward priming Reaction times (ms) Word = yes Word = no Overall Error rates (%) Word = yes Word = no Overall

Related

541 615

Unrelated

568 644

2.8 4.7

557 634

27 28 28

4.0 6.4

576 659

3.1 4.2

M

1.2 1.7 1.5 19 25 22

5.2 5.7

2.1 1.6 1.8

JT* SE

8 7 5 1.5 1.1 0.9 8 9 6 1.0 1.3 0.8

U

27 -28 i

1.2

-1.7

0.3

19 -25 -3 2.1

-1.6

0.3

SE

8 7 7 1.5 1.1 0.9 8 9 7 1.0 1.3 0.9

"Associative priming: Mean of unrelated minus related prime condition; standard errors. Judgmental tendency: Mean of unrelated minus related prime condition (for word = yes) and related minus unrelated prime condition (for word = no); standard errors. tic network account or in a parallel distributed version. But the evaluation task, commonly used in affective priming research, confounds task relevance of the prime and primetarget relation; that is, whenever the valence of the prime is congruent with the valence of the target, it matches the correct response. Therefore, affective priming effects in the evaluation task might be due to interference and, possibly, facilitation processes concerning competing or concurrent pathways from stimulus to response (Wentura, 1997,1999). Second, an affective congruency effect is obtained in the lexical decision task even though congruent and incongruent primes no longer differ with regard to predicting the correct target response. This finding is congruent with the idea that affective congruency effects are produced by spreading activation. Third, however, for affective congruency effects in the lexical decision task, the judgmental tendency model of Klauer (1991; Klauer & Stem, 1992) can be applied. Presentation of prime and target creates a tendency to answer the question "Is (prime) (target)!" (e.g., "Is DEATH wise?"), at least in cases in which primes are nouns and targets are adjectives. Accordingly, congruence or incongruence between prime and target valence triggers an automatic tendency to answer in an affirmative or negative way, respectively. Varying the assignment of yes and no responses to the category of word targets should therefore alter the sign of the congruence effect: A match of the judgmental tendency with the response (i.e., responding with yes in cases of an affective congruent pair or no in cases of an incongruent pair) will lead to faster RTs than a mismatch (i.e., responding with no in cases of an affective congruent pair or yes in cases of an incongruent one). This model was confirmed in Experiment 2.

Fourth, by adding additional assumptions to well-known postlexical accounts of associative relatedness effects in the lexical decision task one can explain the response match effect as well. Because the manipulation of response assignment that was used in Experiment 2 had not yet been used in standard priming experiments, Experiment 3 tested the variation of response assignment with associatively related prime-target pairs. Here, the typical semantic priming effect appeared for word = yes as well as for word = no responses. Moreover, it appeared in a backward priming condition. Whereas this result seems to support the importance of a retrospective semantic matching process (see Neely & Keefe, 1989; Neely et aL, 1989) in explaining associative relatedness effects, it corroborates the claim that affective congruence effects in the lexical decision task are due to a different kind of process—the judgmental tendency. Fifth, however, there remains one argument that seems to reconcile previous postlexical accounts of associative relatedness effects and the judgmental tendency account of affective priming effects. The retrospective semanticmatching theory as well as the compound cue theory assume that if prime-target relatedness provides information about what the correct response to the target is, participants use that information to bias their target response. That is, given the special instructions in Experiments 2 and 3, participants used the prime-target congruence-relatedness to bias a "word'* decision and then translate that decision into the appropriate yes versus no response. Thus, there are three components in this process: the congruence-relatedness of prime and target, the (internal) "word" decision, and the overt yes or no response. Then, response compatibility effects (as those assumed by the judgmental tendency theory) might depend on whether the components are

466

WENTURA

evaluatively laden or valence free. It might now be argued that the "word" decision as well as the prime-target congruence-relatedness both have a positive valence. In accordance with the judgmental tendency account, this congruence will be compatible to a yes response but incompatible to a no response. Then, relatedness effects will depend on the yes versus no response mode as was shown in Experiment 2 (with affectively congruent prime-target pairs) but not in Experiment 3 (with associatively related primetarget pairs). Thus, the results of Experiment 3 eventually falsify nothing but the assumption of a positive connotation of associative prime-target relatedness, whereas the results of Experiment 2 are on first sight compatible with this slightly modified version of the previous postlexical accounts by claiming that affective prime-target congruence is positively valenced. Given this backdrop, the assumption that affective priming effects and semantic priming effects are mediated by entirely different mechanisms cannot be maintained. However, this version predicts two additive effects. First, a positive priming effect results from primetarget congruence biasing a "word" decision. This effect is not moderated by response mode. Second, because of judgmental tendencies, a positive effect will be added for the yes responses, but a negative one will be added for the no responses. Thus, depending on the magnitude of the two effects, the overall priming effect for no responses will be positive, zero, or negative. It will always be less than the overall priming effect for yes responses, but it will not have the same absolute magnitude. This version is disproven by the priming effect for no responses being of the same absolute magnitude as that for yes responses in Experiment 2. The most similar theoretical supposition to the judgmental tendency model was made by West and Stanovich (1982; see also Stanovich & West, 1983) in the context of sentence priming (i.e., paradigms with complete or fragmented sentences as a prime context for target words). They hypothesized that incongruent sentence target pairs will trigger a tendency to negate (in the sense of "This is a wrong statement!"); conversely, congruent pairs might trigger a tendency to affirm. Supposing, as the authors did, that the lexical decision task might be implicitly coded as requiring a yes response to words, responses to incongruent pairs will be slower or more error prone than responses to congruent pairs. De Groot (19&4,1985) transferred this explanation to the standard word prime paradigm and called the supposed mechanism postlexical coherence checking or meaning integration (see Holender, 1992, for an incorporation of those suggestions into a larger framework for explaining congruity effects in response latency paradigms). However, Experiment 3 revealed that response match processes do not seem to contribute to relatedness effects for all kinds of materials. Currently, one possibility is that certain syntactical forms of prime-target pairs might be a prerequisite for response match effects in the lexical decision task: Sentence fragments as primes might create a stance of wondering whether the target word is a correct completion, as West and Stanovich (1982) speculated (see also Goodman, McClelland, & Gibbs, 1981; West & Stanov-

ich, 1982). The same seems to be true for noun-adjective pairs as they fit a schema of ascription. So it might also be the case that affectively neutral noun-adjective pairings with a high degree of semantic relatedness (snow-white, coalblack) cause judgmental tendencies.2 Because semantic relatedness was carefully avoided, this does not undermine the claim that affective congruence or incongruence plays the dominant role in Experiments 2 and 3. On the other hand, the prerequisite of a "schema of ascription" makes it easily understandable that Klauer et al. (1995) found no affective congruence effects with the lexical decision task by using adjective-adjective pairs. What do these results mean for the overall understanding of affective congruence effects? The four tasks that were used in affective priming research can be divided into two sets. For the evaluation task and the lexical decision task, reliable congruence effects were found but can be explained by response compatibilities. For the pronunciation task and the Stroop-priming task, congruence effects could not (Stroop-priming task) or not obviously (pronunciation task; but see Rothermund & Wentura, 1998; Wentura, 1997) be explained by response compatibilities. There were either unequivocal null results (Stroop-priming task) or mixed results (pronunciation task).3 Overall, the results from the present experiments converge on the conclusion that the presentation of an affectively laden to-be-ignored prime has obtrusive effects on the processing of the target. It seems that there is a strong tendency for affective information to "meddle" with the formation of responses. Depending on one's frame of reference, this gives rise to a narrower and yet broader view of affective processing than is suggested by the spreading activation account. According to the spreading activation account, processing of the prime leads to a flow of activation from the prime node via a corresponding affect node to all other nodes whose concepts are evaluatively congruent. Thus, prime presentation alters the state of memory by

2 K.-C. Klauer (personal communication, July, 13,1995) and an anonymous reviewer made this suggestion. 3 Most recently, Klauer and Musch (1997) failed to find an affective priming effect in an extensive series of pronunciation experiments, some of which used procedures highly similar to those of Bargh et al. (1996), with a total sample size of N = 630 (see also De Houwer et al., 1998; Hermans, 1996). Of course, this does not necessarily imply that the former findings of affective priming (Bargh et al., 1996; Hermans et al., 1994) were due to a Type I error. But the results of Klauer and Musch raise serious doubts on the most parsimonious account of the former findings, that is, a spreading activation account Therefore, to explain the discrepancy between the results, we have to consider other possibilities. For instance, given that even in semantic priming research there is some evidence of priming effects that can be attributed only to the production stage of the pronunciation task (Balota, Boland, & Shields, 1989), we might speculate whether there are some basic differences in the way positive and negative words are typically pronounced. If so, affective priming effects in pronunciation might also be due to "response congruency" effects (though of a different kind), as in the lexical decision and affective judgment tasks.


shifting the accessibility of concepts. This process is rather unobtrusive, noticeable at a behavioral level only through changed probabilities of using facilitated and nonfacilitated concepts. Instead, the present data suggest that affective priming effects are based on the independent and automatic extraction of the affective component of both primes and targets. Depending on the task, different processes of facilitation and interference are triggered by these components. In the evaluation task, automatic extraction of prime valence may trigger the corresponding response, thus producing interference with or facilitation of processing the target. In the lexical decision task, however, there is a tendency to judge the prime-target relationship as true or false according to a match or mismatch of valences. In conclusion, to use a spatial metaphor, the prime influences information processing in a narrow "vertical" channel from perception to response. The spreading activation model, however, suggests the view of a broad "horizontal" influence, that is, spreading of activation, solely on a representational level. Whereas the latter view reveals nothing about the special character of affective information processing, the former view finds more agreement with other studies in this domain. Because detecting risks and opportunities is the fundamental function of affective and emotional processing (e.g., Frijda, 1988; Lang, 1995; Simon, 1967), affective stimuli have the power to "meddle" with ongoing processes. One may assume that the valence of stimuli indicates this general relevance for the individual. Information processing is not simply geared toward the acquisition of knowledge but enables the organism to act in an environment full of opportunities and risks. Thus, the "meddling-in" of affective stimuli should cause interference with ongoing processes if these are unrelated to the stimulus (e.g., Pratto & John, 1991; Roskos-Ewoldsen & Fazio, 1992; Wentura, Rothermund, & Bak, in press); "meddling-in" will mean an integration of other information if the affective component is useful in this context (e.g., contexts requiring judgments).

References Anderson, J. R. (1974). Retrieval of propositional information from long-term memory. Cognitive Psychology, 6, 451-474. Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard University Press. Anderson, J. R., & Pirolli, P. L. (1984). Spread of activation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 791-798. Balota, D. A., Boland, J. E., & Shields, L. W. (1989). Priming in pronunciation: Beyond pattern recognition and onset latency. Journal of Memory and Language, 28, 14-36. Balota, D. A., & Chumbley, J. I. (1984). Are lexical decisions a good measure of lexical access? The role of word frequency in the neglected decision stage. Journal of Experimental Psychology: Human Perception and Performance, 10, 340-357. Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the automatic attitude activation effect. Journal of Personality and Social Psychology, 62. 893-912. Bargh, J. A., Chaiken, S., Raymond, P., & Hymes, C. (1996). The automatic evaluation effect: Unconditional automatic attitude

467

activation with a pronunciation task. Journal of Experimental Social Psychology, 32, 104-128. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 125M48. Bower, G. H. (1987). Commentary on mood and memory. Behaviour Research and Therapy, 25,443-455, Bower, G. H. (1991). Mood congruity of social judgments. In J. P. Forgas (Ed,), Emotion and social judgments (pp. 31-53). Oxford, England: Pergamon Press. Bower, G. H., & Cohen, P. R. (1982). Emotional influences in memory and thinking: Data and theory. In M. S. Clark & S. X Fiske (Eds.), Affect and cognition (pp. 291-331). Hillsdale, NJ: Erlbaum. Chaiken, S., & Bargh, J. A. (1993). Occurrence versus moderation of the automatic attitude activation effect: Reply to Fazio. Journal of Personality and Social Psychology, 64, 759-765. Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82, 407428. De Groot, A. M. B. (1984). Primed lexical decision: Combined effects of the proportion of related prime-target pairs and the stimulus-onset asynchrony of prime and target. Quarterly Journal of Experimental Psychology, 36A, 253-280. De Groot, A. M. B. (1985). Word-context effects in word naming and lexical decision. Quarterly Journal of Experimental Psychology, 37A, 281-297. De Houwer, J., & Hermans, D. (1994). Differences in the affective processing of words and pictures. Cognition and Emotion, 8, 1-20. De Houwer, J., Hermans, D., & Eelen, P. (1998). Affective and identity priming with episodically associated stimuli. Cognition and Emotion, 12, 145-169. Draine, S. C , & Greenwald, A. G. (1998). Replicable unconscious semantic priming. Journal of Experimental Psychology: General, 127, 286-303. Fazio, R. H. (1986). How do attitudes guide behavior? In R. M. Sorrentino & E. T. Higgins (Eds.), The handbook of motivation and cognition: Foundations of social behavior (pp. 204-243). New York: Guilford Press. Fazio, R. H. (1993). Variability in the likelihood of automatic attitude activation: Data reanalysis and commentary on Bargh, Chaiken, Govender, and Pratto (1992). Journal of Personality and Social Psychology, 64, 753-758. Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C , & Kardes, F. R. (1986). On the automatic activation of attitudes. Journal of Personality and Social Psychology, 50, 229-238. Fox, E. (1995). Negative priming from ignored distractors in visual selection: A review. Psyckonomic Bulletin &. Review, 2, 145173. Frijda, N. H. (1988). The laws of emotion. American Psychologist, 43, 349-358. Goodman, G. O., McClelland, J. L., & Gibbs, R. W. (1981). The role of syntactic context in word recognition. Memory & Cognition, 9, 580-586. Greenwald, A. G., Draine, S. C , & Abrams, R. L. (1996, September 20). Three cognitive markers of unconscious semantic activation. Science, 273, 1699-1702. Greenwald, A. G., Klinger, M. R., & Liu, T. J. (1989). Unconscious processing of dichoptically masked words. Memory & Cognition, 17, 35-47. Greenwald, A. G., Klinger, M. R., & Schuh, E. S. (1995). Activation by marginally perceptible ("subliminal") stimuli: Dissociation of unconscious from conscious cognition. Journal of Experimental Psychology: General, 124, 22-42. Hager, W., Mecklenbrauker, S., Moller, H., & Westermann, R.

468

WENTURA

(1985). Emotionsgehalt, Bildhaftigkeit, Konkretheit und Bedeutungshaltigkeit von 580 Adjektiven: Ein Beitrag zur Nonnierung und zur Prilfung einiger Zusammenhangshypothesen [Emotionality, imageability, concreteness, and meaningfulness of 580 adjectives: A contribution to norming and test of correlations]. ArchivjUr Psychologie, 137, 75-97. Hasselhorn, M., & Grube, S. (1994). Erstassoziationen von Kindern und Envachsenen zu 53 Substantiven [First-order associations of children and adults to 53 nouns]. In W. Hager & M. Hasselhom (Eds.), Handbuch deutschsprachiger Wortnormen [Handbook of German word norms] (pp. 59-64). Gottingen, Germany: Hogrefe. Hasselhorn, M., & Hager, W. (1994). Assoziationsnormen zur Pruning von Hypothesen tiber ZusammenhSnge zwischen der assoziativen Bedeucungshaltigkeit, der eingeschafzten Bedeutungshaltigkeit und der Bildhaftigkeit [Association norms to test hypotheses concerning relationships between associative meaningfulness, rated meaningfiilness and imageability]. In W. Hager & M. Hasselhorn (Eds.), Handbuch deutschsprachiger Wortnormen [Handbook of German word norms] (pp. 45-53). Gottingen, Germany: Hogrefe. Heathcote, A. (1988). Screen control and timing routines for the IBM microcomputer family using a high-level language. Behavior Research Methods, Instruments, and Computers, 20, 289297. Hermans, D. (1996). Automatische stimulusevaluatie. Een experimentele analyse van de voorwaarden voor evaluatieve stimulusdiscriminatie aan de hand van net affectieve-prirningparadigma, [Automatic stimulus evaluation. An experimental analysis of the preconditions for evaluative stimulus discrimination using an affective priming paradigm]. Unpublished doctoral dissertation, University of Leuven, Belgium. Hermans, D., Baeyens, F., & Eelen, P. (1998). Odours as affectiveprocessing context for word evaluation: A case of cross-modal affective priming. Cognition and Emotion, 12, 601-613. Hermans, D., De Houwer, J., & Eelen, P. (1994). The affective priming effect: Automatic activation of evaluative information in memory. Cognition and Emotion, 8, 515-533. Hermans, D., De Houwer, J., & Eelen, P. (1996). Evaluative decision latencies mediated by induced affective states. Behaviour Research and Therapy, 34,483-488. Hermans, D., Van den Broeck, A., & Eelen, P. (1998). Affective priming using a colour-naming task: A test of an affectivemotivational account of affective priming effects. Zeitschrift fUr Experimented Psychologie, 45, 136-148. Hill, A. B., & Kemp-Wheeler, S. M. (1989). The influence of context on lexical decision times for emotionally aversive words. Current Psychology Research and Reviews, 8, 219-227. Holender, D. (1992). Expectancy effects, congruity effects, and the interpretation of response latency measurement. In J. Alegria, D. Holender, J. Junca de Morais, & M. Radeau (Eds.), Analytic approaches to human cognition (pp. 351-375). Amsterdam: Elsevier. Kahan, T. A., Neely, J. H., & Forsythe, W. J. (1999). Dissociated backward priming effects in lexical decision and pronunciation tasks. Psychonomic Bulletin & Review, 6, 105-110. Kemp-Wheeler, S. M., & Hill, A. B. (1992). Semantic and emotional priming below objective detection threshold. Cognition and Emotion, 6, 113-128. Kitayama, S., & Howard, S. (1994). Affective regulation of perception and comprehension: Amplification and semantic priming. In P. M. Niedenthal & S. Kitayama (Eds.), The heart's eye (pp. 41-65). New York: Academic Press. Klauer, K. C. (1991). Einstellungen. Der Einfiufi der affektiven Komponente aufdas kognitive Urteilen [Attitudes: The influence

of the affective component on cognitive judgments]. Gottingen, Germany: Hogrefe. Klauer, K. C. (1998). Affective Priming. European Review of Social Psychology, 8, 67-103. Klauer, K. C , & Musch, J. (1997). Affective priming: The puzzle of the naming task. Unpublished manuscript, University of Bonn, Bonn, Germany. Klauer, K. C , RoBnagel, C , & Musch, J. (1995). Affective priming effects: Evidence for assimilation and contrast. Unpublished manuscript Klauer, K. C , RoBnagel, C , & Musch, J. (1997). List-context effects in evaluative priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23, 246-255. Klauer, K. C , & Stern, E. (1992). How attitudes guide memorybased judgments: A two-process model. Journal of Experimental Social Psychology, 28, 186-206. Koriat, A. (1981). Semantic facilitation in lexical decision as a function of prime-target association. Memory & Cognition, 9, 587-598. Lang, P. J. (1995). The emotion probe. Studies of motivation and attention. American Psychologist, 50, 372-385. MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109, 163-203. Masson, M. E. J. (1995). A distributed memory model of semantic priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 3-23. May, C. P., Kane, M. I, & Hasher, L. (1995). Determinants of negative priming. Psychological Bulletin, 118, 35-54. McClelland, J. L., Rumelhart, D. R, & PDP Research Group (Eds.). (1986). Parallel distributed processing: Vol. 2. Explorations in the microstructure of cognition. Cambridge, MA: MIT Press. McRae, K., de Sa, V. R., & Seidenberg, M. S. (1997). On the nature and scope of featural representations of word meaning. Journal of Experimental Psychology: General, 126, 99-130. MGUer, H., & Hager, W. (1991). Angenehmheit, Bedeutungshaltigkeit, Bildhaftigkeit und Konkretheit-Abstraktheit von 452 Adjektiven [Pleasantness, meaningfulness, imageability and concreteness of 452 adjectives]. Sprache und Kognition, 10, 39-51. Musch, J., & Klauer, K. C. (1997). Der Anteilseffekt beim affektiven Priming: Replikation und Bewertung einer theoretischen Erklaiung [The relatedness proportion effect in affective priming: Replication and evaluation of a theoretical explanation]. Zeitschrift fiir Experimentelle Psychologie, 44, 266-292. Neely, J. H. (1976). Semantic priming and retrieval from lexical memory: Evidence for facilitatory and inhibitory processes. Memory & Cognition, 4, 648-654. Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limitedcapacity attention. Journal of Experimental Psychology; General, 106, 226-254. Neely, J. H. (1991). Semantic priming effects in visual word recognition: A selective review of current findings and theories. In D. Besner & G. W. Humphreys (Eds.), Basic processes in reading. Visual word recognition (pp. 264-336). Hillsdale, NJ: Erlbaum. Neely, J. H., & Keefe, D. E. (1989). Semantic context effects on visual word processing: A hybrid prospective-retrospective processing theory. The Psychology of Learning and Motivation, 24, 207-249. Neely, J. H., Keefe, D. E., & Ross, K. L. (1989). Semantic priming in the lexical decision task: Roles of prospective primegenerated expectancies and retrospective semantic matching.


Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 1003-1019. Neill, W. X, Valdes, L. A., & Terry, K. M. (1995). Selective attention and the inhibitory control of cognition. In F. N. Dempster & C. J. Brainerd (Eds.). Interference and inhibition in cognition (pp. 207-261). San Diego, CA: Academic Press. Peterson, R. P., & Simpson, G. B. (1989). Effect of backward priming on word recognition in single-word and sentence contexts. Journal of Experimental Psychology; Learning, Memory, and Cognition, 15, 1020-1032. Pollatsek, A., & Well, A. D. (1995). On the use of counterbalanced designs in cognitive research: A suggestion for a better and more powerful analysis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 785-794. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information processing and cognition: The Loyola symposium (pp. 55-85). Hillsdale, NJ: Erlbaum. Pratto, F., & John, O. P. (1991). Automatic vigilance: The attention-grabbing power of negative social information. Journal of Personality and Social Psychology, 61, 380-391. Ratcliff, R., & McKoon, G. (1988). A retrieval theory of priming in memory. Psychological Review, 95, 385-408. Riedlinger, C. B. (1994). Wortassoziationsnormen fur 80 konkrete deutsche Reizworter [Word association norms for 80 concrete German stimulus words]. In W. Hager & M. Hasselhom (Eds.), Handbuch deutschsprachiger Wortnormen [Handbook of German word norms] (pp. 54-58). Gottingen, Germany: Hogrefe. Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992), On the orienting value of attitudes: Attitude accessibility as a determinant of an object's attraction of visual attention. Journal of Personality and Social Psychology, 63, 198-211. Rothermund, K., & Wentura, D. (1998). Ein fairer Test fur die Aktivationsausbreitungshypothese: affektives Priming in der Stroop-Aufgabe [An unbiased test of a spreading activation account of affective priming: Analysis of affective congruency effects in the Stroop task]. Zeitschrift fur Experimentelle Psychologic, 45, 120-135. Rumelhart, D. E., McClelland, J. L., & PDP Research Group (Eds.). (1986). Parallel distributed processing: Vol. 1. Explorations in the microstructure of cognition. Cambridge, MA: MIT Press. Schmuck, R (1994). Restringieite Assoziationsbefragung zu 56 Begriffen der Alltagssprache [Restricted association interviews to 56 concepts of colloquial language]. In W. Hager & M. Hasselhorn (Eds.), Handbuch deutschsprachiger Wortnormen [Handbook of German word norms] (pp. 70-76). Gottingen, Germany: Hogrefe.

469

Seidenberg, M. S., Waters, G. S., Sanders, M., & Langer, P. (1984). Pre- and postlexical loci of contextual effects on word recognition. Memory & Cognition, 12, 315-328. Simon, H. A. (1967). Motivational and emotional controls of cognition. Psychological Review, 74, 29-39. Stanovich, K. E., & West, R. F. (1983). On priming by a sentence context. Journal of Experimental Psychology: General, 112, 1-36. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-662. Tukey, J. W. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley. Warren, R. E. (1972). Stimulus encoding and memory. Journal of Experimental Psychology, 94, 90-100. Warren, R. E. (1974). Association, directionality, and stimulus encoding. Journal of Experimental Psychology, 102, 151-158. Wentura, D. (1994). Gibt es ein "affektives Priming" im semantischen Ged&chtnis? [Is there an affective priming effect in memory?] (Reports of the Psychological Institute IV). MUnster, Germany: University of MUnster. Wentura, D. (1997). Zur mentalen Representation affektivevaluativer Komponenten: die Netzwerkmetapher und das Paradigma des affektiven Primings [The mental representation of affective-evaluative components: the network metaphor and the "affective priming" paradigm]. In H. Mandl (Ed.), Bericht fiber den 40. Kongrefi der Deutschen Gesellschaft fur Psychologic in MUnchen 1996 (pp. 446-453). Gottingen, Germany: Hogrefe. Wentura, D. (1998). Affektives Priming in der Wortentscheidungsaufgabe: Evidenz fur postlexikalische Urteilstendenzen [Affective priming in the lexical decision task: Evidence for postlexical judgment tendencies]. Sprache und Kognition, 17, 125137. Wentura, D. (1999). Activation arid inhibition of affective information: Evidence for negative priming in the evaluation task. Cognition and Emotion, 13, 65-91. Wentura, D., Rothermund, K., & Bak, P. (in press). Automatic vigilance: The attention grabbing power of approach- and avoidance-related social information. Journal of Personality and Social Psychology. West, R. F., & Stanovich, K. E. (1982). Source of inhibition in experiments on the effect of sentence context on word recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 385-399.

Received August 12, 1996 Revision received July 2,1999 Accepted July 20, 1999 •