awareness in negative priming tasks Running head: AWARENESS IN NEGATIVE PRIMING PARADIGMS

Negative priming with masked distractor-only prime trials: Awareness moderates Negative Priming

Christian Frings & Dirk Wentura

Saarland University

Christian Frings Saarland University Faculty of Behavioral Sciences Department of Psychology Building 1 P.O. Box 15 11 50 D-66041 Saarbrcken c.frings@mx.uni-saarland.de

-------- Experimental Psychology, in press -----------

awareness in negative priming tasks Abstract The literature yields inconsistent evidence for negative priming (NP) following masked distractor-only prime trials. We contrast two different hypotheses on the inconsistent findings, one which is most compatible with the temporal discrimination theory that relates the sign of priming effects to the absence vs. presence of prime awareness and one

which is most compatible with the inhibition and episodic retrieval accounts that relates the sign of priming effects to the prime event being categorized as a to-be-attended vs. to-beignored event. In two experiments it turned out that participants awareness of the masked stimuli caused the different results (with participants being not aware of the primes showing NP), whereas the factor prime color = probe target color vs. prime color = probe distractor color (i.e., the prime contains the to-be-attended vs. the to-be-ignored signal) did not moderate NP. These findings are discussed with regard to theories of negative priming and the debate on conscious versus unconscious perception.

awareness in negative priming tasks Negative priming with masked distractor-only prime trials: Awareness moderates Negative Priming Since its introduction by Dalrymple-Alford and Budayr (1966), Neill (1977), and

Tipper (1985), the Negative Priming (NP) phenomenon has been a widely researched topic in cognitive psychology. The original finding was that responses to a previously ignored stimulus were retarded if the same stimulus was subsequently repeated as a target. Usually, this effect is demonstrated in selective attention tasks, where target stimuli are accompanied by distractor-stimuli. Thus, if a stimulus is to be ignored in the first trial (the prime trial) and then presented as the target in the consecutive trial (the probe; this condition is called the ignored-repetition condition) responses are slower compared to a control condition with unrepeated stimuli (i.e., the unrepeated condition). Despite extensive evidence in favor of the NP-effect, the underlying mechanisms continue to be hotly debated today. The two most popular theories about NP are the inhibition account (e.g., Tipper, 1985; Houghton & Tipper, 1994; Tipper, 2001) and the episodic retrieval account (e.g., Neill & Valdes, 1996). The inhibition account focuses on active inhibition of the prime-distractor during the process of selecting the prime target; inhibition is assessed by presenting prime distractors as probetargets in the consecutive trial. The episodic retrieval account highlights the ability of a stimulus to retrieve its latest appearance: Thus, if a probe target retrieves its latest appearance as the prime distractor, a do-not-response tag will hamper response generation. Recently, Milliken, Joordens, Merikle, and Seifert (1998) developed an alternative perspective on NP, the temporal discrimination theory. They assume a comparison process that detects discontinuities or mismatches between the mental representation of the prime trial and the perception of the probe trial. If the quality of match is high as in the case of the prime target being identical to the probe target (i.e., the attended repetition condition of NP experiments), no mismatch is detected and any response facilitation due to the fact that the same stimulus was just processed before will lower response time (compared to the

awareness in negative priming tasks unrepeated condition). If the quality of match is poor as in the case of the unrepeated condition of NP experiments, a quick mismatch detection occurs, immediately starting a process of generating a response to the probe target. If, however, the quality of match is intermediate as in the case of the ignored repetition condition of NP experiments the mismatch process needs more time to arrive at the final mismatch decision such that the process of generating a response to the probe target is delayed compared to the unrepeated condition, resulting in a NP-effect. NP as a function of prime awareness In this regard, one important finding was that NP even occurs after distractor-only prime trials, that is, even without a selection process in the prime-trial (Milliken et al., 1998;

see also Milliken & Joordens, 1996). In an experimental series (Milliken et al. 1998, Exp. 2a2c & Exp. 3) a distractor was presented very briefly (33 ms) during the prime trial and was masked by a forward and backward mask. Nevertheless, NP was observed when the probe trial contained a distractor. These results fitted with predictions of temporal discrimination theory: Probe targets could neither be identified as old nor as new if they immediately followed masked and briefly displayed distractors; thus, a delayed response to the probe target will occur and cause NP. Furthermore, participants in Milliken et al.s experiments stated that they were not aware of the masked primes. Recently, Healy and Burt (2003) yielded further evidence for this masked NP-effect. They found slower responses to targets in a Stroop-color naming task if the same stimuli were directly before the Stroop-display briefly presented and followed by a backward mask. These results are interesting because the findings of Milliken et al. (1998) and Healy and Burt (2003) contrast findings of similar experiments by Allport, Tipper and Chmiel (1985, Exp. 4 & 5). In these experiments Allport et al. (1985) varied the SOA (stimulus onset asynchrony) of the prime display and backward mask. For short SOAs participants of Allport and colleagues yielded a positive priming effect whereas for long SOAs NP occurred. As in the experiments of Milliken and colleagues, participants

awareness in negative priming tasks seemed not to be fully aware of the prime stimuli (their identify-accuracy of masked stimuli was not better than chance). Allport et al. concluded that NP depends on selection of a target

against a distractor. Thus, if a prime display is presented only at short SOAs no NP will occur because participants could not select a target against a distractor. Milliken et al., however, argued that the masked distractors in their experiments can not be identified as old or as new because the quality of match of a masked distractor is intermediate (see above). For a new theoretical approach to NP it seems important to elaborate on this inconsistency1. However, the literature yields even more puzzling findings for masked NP. Recently, in an attempt to replicate the distractor-only NP effect Neill and Kahan (1999, Exp. 1a & Exp. 1b) reported inconsistent results with masked distractor-only-prime trials. In fact, in their Experiment 1b they found as one would expect from the broader literature on priming positive priming effects following distractor-only prime trials with masked distractors (this result parallels the findings of Allport et al., 1985) . In contrast, in their Experiment 1a (which was essentially the same as Experiment 1b) they replicated the findings of Milliken et al. (1998). Neill and Kahan (1999, p. 306) noted post hoc that the participants awareness of the masked primes might be the moderator of their inconsistent results: Awareness of the primedistractor might be a predictor of quality of match to the probe target. However, neither Milliken and colleagues (1998) nor Neill and Kahan (1999) nor Healy and Burt (2003) used sensitive procedures to measure participants awareness of the prime distractors. Because this phenomenon is quite central for the temporal discrimination theory, and because the NP literature needs further research to clarify these inconsistencies, with Experiment 1 and 2 we tried to replicate NP following masked distractor-only prime trials while using direct tests to measure participants awareness of the masked stimuli (see, e.g., Greenwald, Klinger, & Schuh, 1995). Our direct test allows for a more conservative test of participants awareness (see, e.g., Merikle, Smilek, & Eastwood, 2001). If Neil and Kahan are correct with their post hoc argument, the number of participants being aware of the single

awareness in negative priming tasks prime distractor will predict whether a negative, a null, or a positive priming effect will be observed. Moreover, our direct test allows for a split of the sample with regard to the prime distractors visibility. NP as a function of prime status There might be, however, a further argument to explain the diverging results of Neil and Kahans (1998) Experiments 1a and 1b, ironically one that helps to bring the inhibition account and episodic retrieval theory again into play. The prime distractors in their experiments (as well as those of Milliken et al., 1998, and ours in Exp. 1) were of neutral color (i.e., they neither matched the probe-target nor the probe-distractor in color). One can

argue that the prime distractor is therefore not specified as a to-be-ignored or a to-be-attended object because color was used for target-distractor discrimination in the probe trial. The inhibition account assumes that a to-be-ignored stimulus can be identified by the feature of, e.g., color, location, temporal sequence etc. Thus, if the probe display contains color as the differentiating feature, one can argue that the prime of neutral color is neither specified as a stimulus that should be attended-to nor as one that should be ignored. Likewise, the episodic retrieval theory assumes that distractors are marked by a do-not-response tag. Distractors of the probe display, however, can only be identified by color. Thus, one can argue that a prime of neutral color is neither specified as a stimulus that should be tagged nor as one that should not be tagged. In both cases it is at the discretion of the participants to process the prime in either the one way or the other. Thus, both the inhibition account and the episodic retrieval theory predict that the number of participants who treat the prime distractor as a to-be-attended object might determine whether the NP effect turns into a null or even a positive priming effect. This argument fits with the observation of Neill and Kahan that part of the sample of their Experiment 1b (i.e., the one resulting in a positive priming effect) had previously participated in other experiments in which they were instructed to attempt to identify a

awareness in negative priming tasks masked word (p. 306). This experience might have lead to adopting the stance that the brief flicker (i.e., the prime distractor) is a to-be-attended event. Moreover, this argument is also compatible with regard to the procedure of Allport and colleagues (1985). In their

experiments participants individual thresholds for masked stimuli were surveyed before they worked through the NP task. Therefore, participants were fully aware that masked stimuli were presented, and they should try to identify them, that is, they handled them as to-beattended objects. Thus, in Experiment 2, we varied the color of the prime distractor, such that it either matched that of the probe target or that of the probe distractor. In the former case the prime distractor carries the to-be-attended signal, in the latter case the to-be-ignored signal. Given the backdrop of the inhibition and episodic retrieval accounts, we would predict positive priming after prime distractors which match the probe target in color whereas negative priming is predicted after prime distractors which match the color of the probe distractor. Both accounts are not very specific about the role of awareness. Thus, if color match but not awareness will be the dominant predictor for the sign of the priming effect, we can take that as a piece of evidence for the inhibition and episodic retrieval accounts. However, we should hasten to add that for two reasons the temporal discrimination account is not incompatible to a dependency of the NP effect on color match. Firstly, one can argue that the quality of match is higher in the case of a color match of prime and probe target. However, because the theory assumes a curvilinear relationship of quality of match and response times, the prediction is not clearly specified. Secondly, Milliken et al. (1998) demonstrated that even for supraliminal distractor-only-prime trials a NP effect emerged if participants were instructed to process clearly visible distractors as to-be-ignored-objects whereas a positive priming effect emerged if participants were instructed to process them as to-be-attended objects. Given this, the theory would be compatible with the finding of a positive priming effect in the case of a color match and a NP effect in the case of a mismatch.

awareness in negative priming tasks However, the temporal discrimination account would put more weight on awareness of the prime (as a closer associate of quality of match) than on features that signal relevance or

irrelevance of the prime. In conclusion, finding a dependence of the sign of the priming effect on awareness but not on color match would be most compatible to the temporal discrimination account. Experiment 1 Experiment 1 is primarily concerned with the replication of the masked distractor-only NP effect while assessing participants awareness in a more controlled way. However, we added a further condition. As mentioned above, a single visible prime distractor is not in itself enough to predict positive priming effects. According to Milliken and colleagues (1998), the absence of a strategic component seems to be important. Instructing their participants to willingly ignore the prime distractors, they found NP-effects even if a single distractor stimulus was clearly visible (i.e., it was presented for 200 ms without a mask; Exp. 4). Up to now, masked and strategic ignorance NP effects have never been tested for in a single experiment, using a within-participants design (nevertheless, it is argued that the same process caused these NP-effects). Method Participants. Seventeen undergraduate students (13 women and 4 men) of the University of Jena (aged 19 to 27; M = 23) participated either for partial course credits or were paid an amount of four . Materials. The stimuli set comprised the following high-frequency German nouns: Palme (palm), Perle (pearl), Pulver (powder), Diener (servant), Dosis (dose rate), Donner (thunder), Bogen (bow), Buche (beech), Becher (cup),Teller (plate) , Tenor (tenor), Tunnel (channel). The individual letters of the words had a size of 9 x 5 mm. Distractors in the probe display were always presented in green color whereas probe targets were presented in red color. Both words were presented in uppercase and at close quarters in the center of the

awareness in negative priming tasks screen, one above the other. Half of the targets were presented above the distractor, the other half below (see Design). The distractors of the prime display were always presented in black color.

Design. Essentially, the design comprised two within-subjects factors: (1) presentation modus of prime distractor (subliminal vs. supraliminal) and (2) prime condition (repeated vs. unrepeated). Additionally, the positions of probe distractor and target were counterbalanced. As a between-subjects factor sequence of presentation modus (subliminal presentation first vs. second) was counterbalanced. Procedure. All participants were tested individually. The experiment was conducted on a standard PC with a standard monitor with a resolution of 800 x 600 pixels and a refresh rate of 75 hz. The experiment was conducted with the Inquisit-software (inquisit 1.33) and a Labtec head-set microphone. The participant started each trial by pressing the space bar. Following the practice block, two blocks (one subliminal and one supraliminal) consisting of 120 trials each were presented (half the trials were repeated, half the trials were unrepeated). Roughly following Milliken and colleagues (1998) the sequence of events for the subliminal condition was as follows: First, a fixation marker was presented for 390 ms. Then the premask consisting of 14 @-symbols was presented for 520 ms before it was overwritten by the prime distractor. The prime distractor appeared for 39 ms before the postmask (consisting of 14 @symbols as well) was shown for 520 ms. Participants were instructed to focus their attention on the location of these strings because they would indicate the position of the probe stimuli while simultaneously the content of the symbols was irrelevant to them. On appearance of the probe display, participants pronounced the red word as quickly and as accurately as possible (naming task). The probe display lasted until participants responded to it or until 5000 milliseconds have passed by. In the supraliminal condition the prime distractor was not masked and was presented for 300 ms before it was overwritten by a blank screen which lasted for 520 ms. Most importantly, in the supraliminal condition participants were instructed

awareness in negative priming tasks 10 to willingly ignore the prime distractor. The experimenter coded the correctness of responses of the participant in each trial via another computer. For each trial, three different words were randomly chosen from the stimulus list and were assigned the roles of prime distractor, probe target, and probe distractor. A stimulus was never repeated in a subsequent trial and was not chosen again before all remaining stimuli of the list were selected. Repeated trials were created by replacing the chosen prime distractor with the probe target. Following the 240 experimental trials, participants were told that the flicker (in the subliminal condition) presented before the probe (i.e., the mask-distractor-mask sequence) contained a word. They were instructed to work through 24 prime identification trials to obtain a direct effect of prime awareness. During this block, each stimulus was presented twice as a prime distractor, using the same parameters as before in the subliminal condition. However, the probe display now contained only one word out of the stimulus list that was either identical to the prime distractor or not. Participants had to decide by key-pressing whether probe and prime matched (which was the case in half of the trials) or not. Results A significance level of = .05 was chosen for all analyses throughout the text. Prime awareness. With the data of the direct test, we computed the non-parametric signal detection sensitivity index A (Pollack, 1970; see also Goschke & Kuhl, 1993) with hits being correctly identified words and false alarms being incorrectly identified words.2 Mean A was M = .53 (SD = .12) and was not significantly different from 0.50, t(16) = 1.13, p = .28, ns. Thus, on average, participants had no access to the lexical status of the prime. Additionally, none of them showed a significant contingency between presentation (i.e., whether prime and probe matched or matched not) and response (yes vs. no), all 2 < 1.51, p > .22. Negative Priming. Correct RTs for each participant and RTs that were above 200 ms but below 1700 ms were included in the analysis. This procedure resulted in the elimination

awareness in negative priming tasks 11 of 3% of all trials (error rate was < 1% and hence we did not analyze errors). The betweenparticipants factor sequence of blocks showed neither a main effect nor an interaction with priming, all Fs < 1.50, ns. The factor was therefore discarded in the following analyses. Mean RTs are shown in Table 1. A 2 (presentation condition) x 2 (prime condition) ANOVA with mean RT as the dependent variable yielded a significant main effect for prime condition, F(1,16) = 9.92, p < .01. On average, participants were M = 8 ms (SD = 10 ms; d = .80) slower following repeated prime distractors, thus indicating a significant NP effect. NP did not depend on the prime distractor being presented subliminally or (to be ignored) supraliminally, as is shown by the lack of an interaction effect of prime and presentation conditions, F(1,16) = .09, p = .77, ns. (There was no main effect of presentation, F[1,16] < 1, ns.). Discussion The results clearly replicate the finding of Milliken and colleagues (1998) and are also in line with the results of Experiment 1a of Neill and Kahan (1999), and Experiment 1 of Healy and Burt (2003). Furthermore, the results are consistent with the suggestions of Neill and Kahan (1999) who argued that their positive priming effect (found in their Exp. 1b with masked priming) might be due to participants awareness of masked primes. Here, we found a NP-effect with a proven absence of prime awareness for all participants. As an aside, the results of this experiment lend support to the claim that the same mechanism is at work given subliminal presentation conditions as well as given instructions to willingly ignore supraliminally presented distractors (in the studies of Milliken and colleagues, 1998, this was only varied between experiments). This strengthens the arguments raised by Milliken et al. (1998) for a temporal discrimination theory on NP. Of course, just because there were no participants being aware of the masked primes, our results are not a clear test of the hypothesis that awareness moderates this NP-effect. One further point should be noted. The overall reaction time seems to be higher than in the experiments of Milliken et al. (1998). Thus, with Experiment 2, we decided to try to

awareness in negative priming tasks 12 conceptually replicate the masked NP-effect with modest modifications to Experiment 1, which make Experiment 2, however, more similar to the experiments of Milliken et al. (1998) and Neil and Kahan (1999). For example, probe distractors and probe targets were now presented with interleaved letters. It might be that these differences lead to a higher portion of participants who are aware of the prime distractors. Moreover, a self-constructed software and voicekey-apparatus was used which were designed especially for NP-experiments. It might be that the longer reaction times were caused by technical reasons. Additionally, we tested the hypothesis whether variations of the color of the prime distractor could induce a to-be-attended or a to-be-ignored signal. As argued above (see Introduction),if color match but not awareness would be the dominant predictor for the sign of the priming effect, we can take that as a piece of evidence in favour of the inhibition and episodic retrieval accounts. However, finding a dependence of the sign of the priming effect on awareness but not on color match would be most compatible to the temporal discrimination account. Experiment 2 Method Participants. Twenty undergraduate students (16 women and 4 men) of the University of Jena (aged 19 to 26; M = 23) participated either for partial course credits or were paid an amount of 3 . Design, Materials, and Procedure. Design, materials, and procedure were essentially the same as in Experiment 1 with the following exceptions. A self-constructed software and voicekey-apparatus was used. Presentation of prime distractor (28 ms) now was masked for all participants. Besides manipulating the prime condition (repeated vs. unrepeated), the color of the prime distractor was varied as a within-subjects factor. That is, the prime distractor had either the same color (red or blue) as the probe target or the probe distractor. Whether red or blue color served as the target color was balanced between participants. The individual letters

awareness in negative priming tasks 13 of the words measured 10 x 6 mm. In line with Milliken et al. (1998) and Neil and Kahan (1999), the probe display now contained two interleaved uppercase words in the middle of the screen, one above the other. Following the voice key task participants were told that the flicker before the probe displays had contained a word. Then they had to work through 48 prime identifications trials. During these trials the prime distractors were presented exactly as in the voice key task. However, participants now had to decide whether the flicker contained a word or a non-word by pressing a key. Non-words were created by modifying words from the stimulus set (e.g., PALME was modified to PLIME). In half of the trials the flicker contained a word from the stimulus set, in the other half the flicker contained a non-word. Each word of the stimulus set as well as each non-word was presented two times, once in red and once in blue. Results Prime awareness. The mean non-parametric signal detection sensitivity index A was M = .56 (SD = .19) and not significantly different from chance performance (i.e., .50), t(19) = -1.32, p = .20, ns. However, a closer inspection of the distribution revealed that four participants showed a significant contingency between word identification and response, all 2 > 5.59, p < .05 (A was above .82). Thus, these participants were aware of the subliminally presented prime distractors. For the other participants the masking of the prime distractors was successful, all individual 2 < 2.75, p > .09, ns (A was in the range of .27 to .75). Therefore, we could assume that these participants had no access to the lexical status of prime distractors (in the following called unaware). A preliminary 2 (prime condition) x 2 (primeprobe match) x 2 (prime awareness) ANOVA yielded a significant interaction effect of prime condition and prime awareness, F(1,18) = 8.93, p < .01. Thus, the following analyses are dominantly conducted for the subsample (N = 16) of unaware participants. Of course, for reasons of comparison, we will return to the subsample (N = 4) of aware participants as

awareness in negative priming tasks 14 well. Additionally, color of the masked stimuli had no influence on detection performance, for both (red and blue) distractor conditions |t|(19) < 1, ns. Negative Priming. Correct RTs for each participant and RTs that were above 200 ms but below 1700 ms were included in the analysis. This procedure resulted in the elimination of 1% of the RTs (error rate was < 1% and hence we did not analyze errors). The factors position of probe target (above or below the probe distractor) and color of prime distractor (blue or red) showed no main effect, all Fs < 1.16, ns, nor did they interact with any other condition, all Fs < 2.52, all ps > .13, ns. Thus, they were discarded for further analyses. Table 2 shows the mean reaction times for the prime condition (repeated vs. unrepeated) and the color match (prime distractor having the probe target color or not). For unaware participants, a 2 (prime condition) x 2 (color match) ANOVA yielded a main effect for prime condition, F(1,15) = 7.01, p < .05. Neither the main effect for color match nor the interaction between prime condition and color match was significant, F(1,15) = .48, p = .50, ns, and F(1,15) = .18, p = .68, ns, respectively. Thus, the prime-probe color match had no influence on the NP effect. For both prime-probe match conditions the average NP effect was M = -12 ms (SD = 19 ms; d = .66), that is, participants were on average 12 ms slower after repeated prime distractors regardless of whether the prime distractor had the same color as the probe target or the probe distractor. Both means were significantly different from zero, t(15) = -2.52, p < .05, for the trials with the prime distractor having the same color as the probe target and t(15) = -1.76, p < .05 (one-tailed), for the trials with the prime distractor having a different color as the probe distractor. In contrast, the subsample (N = 4) of aware participants showed on average a positive priming effect of M = 16 ms (SD = 6 ms) for repeated prime distractors that was significantly different from zero, t(3) = 5.58, p < .05. Discussion Again, the results confirm the finding of Milliken et al.(1998) that NP occurs after distractor-only prime trials. More interestingly, Experiment 2 provides evidence for the

awareness in negative priming tasks 15 hypothesis that the contrary results of Neill and Kahan's (1999) experiments 1a and 1b were due to differences in participants awareness. In our Experiment 2 the participants split into two subsamples. The subsample of unaware participants showed a significant NP effect whereas the subsample of aware participants showed a positive priming effect. In contrast to our Experiment 1 and to Experiment 4 of Milliken et al. (1998), the subsample of aware participants can marginally perceive the prime-distractors but lack the intention to actively ignore them. Thus, a NP effect cannot be expected for them. Furthermore, the color of the prime-distractor had no influence on the NP effect. We gave the prime-distractors a clear to-be-attended or to-be-ignored signal, that is, the color of the probe target or the probe distractor. However, it seems that participants do not use this signal in processing the distractor. If they would have used it in the way predicted by the inhibition and episodic retrieval accounts, the experiments with distractor-only prime trials would have lost their power as an element in favour of the temporal discrimination account and as a pinprick against the other theories. General discussion In two experiments we yielded further evidence for a specific NP-effect, namely NP following masked distractor-only prime trials. The reassurance of this finding strengthens the arguments raised by Milliken et al. (1998) and thus could be interpreted as further evidence for the temporal discrimination theory. Furthermore, the inconsistent results on masked NP as reported by different researchers can now be explained with regard to our experiments. As mentioned above, Milliken et al. (1998) reported NP after masked distractor-only-prime trials, as did Neill and Kahan (1999; Exp. 1a), as well as Healy and Burt (2003; Exp. 1) whereas Neill and Kahan (1999; Exp. 1b) as well as Allport et al. (1985; Exp. 4-5) found positive priming after masked prime trials. Against the background of our results, it seems plausible that participants awareness has caused these inconsistencies: Participants must be unaware

awareness in negative priming tasks 16 of masked prime-distractors to show a NP effect. For participants who are aware of masked prime-distractors and lack the intention to ignore them, a positive priming effect will emerge. Obviously, our results add further data to the debate on perception with and without awareness (see, e.g., Merikle et al., 2001). One approach used to demonstrate that the distinction between conscious and unconscious information processing is a meaningful one is to show that participants can produce qualitatively different patterns of performance if they are consciously aware versus unaware (see, e.g., Joordens & Merikle, 1992). Moreover, if participants performance is moderated due to their level of awareness which was measured in a discrimination test this result can be interpreted as a validation of the adequateness of the discrimination test (Marcel, 1980). In our Experiment 2, we found qualitative different patterns of results in dependence of participants` awareness; this suggests that our discrimination test was an adequate measure of awareness and yields further evidence for the dissociation of conscious versus unconscious information processing: Four participants had access to the lexical state of the masked prime distractors (as their performance in the direct test indicated). These participants seemed to have used the consciously processed information of masked stimuli, that is, they may have learned that in half the trials the masked words were directly repeated as the target. Knowing this contingency between prime and probe display, it seems plausible that the participants had intentionally focussed their attention on the prime distractors (and tried to identify them) and had handled them not as distractors but rather as targets. This would explain why a positive priming effect emerged for them. In contrast, participants who could not categorize masked stimuli during the direct test showed a NPeffect. As Merikle and colleagues (2001) stated, information perceived without awareness leads to more automatic reactions that cannot be controlled by the perceiver. Thus, the perceived information of the prime display interfered with processing of the probe display: The probe target could not be identified as new because participants have perceived the probe target directly before the probe display (albeit without awareness) and it could not be

awareness in negative priming tasks 17 identified as old because it was not perceived with awareness. Participants could not intentionally use the information of the prime distractors to control their reactions to the probe display because they had no explicit or conscious access to the prime distractors. Overall, our data can be interpreted as an example for a qualitative difference in performance due to differences in perceiving with or without awareness. Additionally, our results have further implications for the theories of NP. We found that the variation of prime color in our Experiment 2 (i.e., it either matches the color of the probe target or the probe distractor) does not moderate the NP effect. This result has some implications for the inhibition account (Houghton & Tipper, 1994). This model would predict that prime stimuli will be processed with regard to their color. Thus, prime distractors with the color of the probe target should not be inhibited (because the target-color should attract attention) whereas prime distractors with the color of the probe distractor should be inhibited (because the distractor-color would be the signal for ignoring). The inhibition model would therefore predict that for trials with the same color of prime distractor and probe target a positive priming effect will emerge whereas for trials with a different color of prime distractor and probe target a negative priming effect will emerge; our results do not suggest such an interaction, however. 3 The same is true for episodic retrieval theory (Neill & Valdes, 1996). Similarly to the inhibition account, episodic retrieval theory would predict that our colorvariation should moderate the NP-effect. A distractor that is specified as a to-be-attended signal should not be encoded with a do-not-response tag. Thus, episodic retrieval would also predict, that for trials in which color of prime distractor matches color of probe target, no NP emerges. However, although we interpret the results as being most compatible to the temporal discrimination theory, they are still somewhat problematic. The theory assumes a comparison process that detects discontinuities or mismatches between the mental representation of the prime trial and the perception of the probe trial. In Experiment 2, the color of the prime-

awareness in negative priming tasks 18 distractor had either the color of the probe-target or not. This variation should influence the quality of match of prime and probe. However, apparently this was not the case. There are two possible explanations. First, since Milliken et al. (1998) assume a curvilinear relationship between quality of match and response latencies, we can assume that the qualities of match in the color match and color mismatch conditions, respectively, mark two points on the curve, one on the ascending and the other on the descending part of the curve, with comparable yaxis values, i.e., comparable RT-predictions. Second, we might assume that color is a marginal feature that does not alter the quality of match. Whatever the reason might be, it seems that the comparison process is not yet fully understood and needs a more accurate specification. Further research about the temporal discrimination theory should therefore concentrate on the comparison process of prime and probe, and analyse its impact on the NP effect. Thus we can strongly suggest that there is now strong evidence for a specific NP effect, namely NP following masked distractor-only prime trials. The most suitable approach to this NP seems to be the temporal discrimination account. However, it would be crucial for this theory to further analyse the relationship between prime-probe comparison and NP. Furthermore, inconsistent results in the literature on masked NP can now be explained as being due to differences in level of awareness.

awareness in negative priming tasks 19 References Allport, D. A., Tipper, S. P., & Chmiel, N. R. J. (1985). Perceptual integration and postcategorial filtering. In M. I. Posner & O. S. M. Marin (Eds.), Attention and performance XI (pp. 107-132). Hillsdale, NJ: Erlbaum. Dalrymple-Alford, E. C., & Budayr, B. (1966). Examination of some aspects of the Stroop color-word test. Perceptual and Motor Skills, 23, 1211-1214. Greenwald, A. G., Klinger, M. R., & Schuh E. S. (1995). Activation by marginally perceptible (subliminal) stimuli: dissociation of unconscious and conscious cognition. Journal of Experimental Psychology: General, 124, 22-42. Goschke, T., & Kuhl, J. (1993). Representation of intentions: persisting activation in memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19,1211-1226. Healy, D., & Burt, J. S. (2003). Attending to the distractor and old/new discriminations in negative priming. Quarterly Journal of Experimental Psychology, 56A, 421-443. Houghton, G., & Tipper, S. P. (1994). A model of inhibitory mechanisms in selective attention. In D. Dagenbach & T. Carr (Eds.), Inhibitory processes in attention, memory, and language (pp. 53-112). Orlando, FL: Academic Press. Joordens, S., & Merikle, P. M. (1992). False recognition and perception without awareness. Memory & Cognition, 20, 151-159. Marcel, A. J. (1980). Conscious and preconscious recognition of polysemous words: Locating the selective effects of prior verbal context. In R. S. Nickerson (Ed.), Attention and Performance VIII (pp. 435-457). Hillsdale, NJ: Erlbaum. Merikle, P. M., Smilek, D., & Eastwood, J. D. (2001). Perception without awareness. Perspectives from cognitive psychology. Cognition, 79, 115-134. Milliken, B., & Joordens, S. (1996). Negative priming without overt prime selection. Canadian Journal of Experimental Psychology, 50, 333-346.

awareness in negative priming tasks 20 Milliken, B., Joordens, S., Merikle, P., & Seifert, A. (1998). Selective attention: A reevaluation of the implications of negative priming. Psychological Review, 105, 203-229. Neill, W. T. (1977). Inhibition and facilitation processes in selective attention. Journal of Experimental Psychology: Human Perception and Performance, 3, 444-450. Neill, W. T., & Valdes, L. A. (1996). Facilitatory and inhibitory aspects of attention. In A. F. Kramer, M. Coles, & G. D. Logan (Eds.), Converging operations in the study of visual selective attention (pp. 77-106). Washington, DC: American Psychological Association. Neill, W.T., & Kahan, T. A. (1999). Response conflict reverses priming: A replication. Psychonomic Bulletin & Review, 6, 304-308. Pollack, I. (1970). A nonparacmetric procedure for evaluation of true and false positives. Behavior Research Methods and Instrumentation, 2, 155-156. Tipper, S. P. (1985). The negative priming effect: Inhibitory effects of ignored primes. Quarterly Journal of Experimental Psychology, 37A, 571-590. Tipper, S. P. (2001). Does negative priming reflect inhibitory mechanisms? A review and integration of conflicting views. Quarterly Journal of Experimental Psychology, 54A, 321343.

awareness in negative priming tasks 21 Author note Christian Frings and Dirk Wentura, Department of Psychology, Saarland University, Germany. Correspondence concerning this article should be addressed to Christian Frings, Saarland University, Faculty of Behavioral Sciences, Department of Psychology, Building 1, P.O. Box 15 11 50, D-66041 Saarbrcken, Germany or via email to c.frings@mx.uni-saarland.de. The authors would like to thank Jrg Peuckert for his technical support.

awareness in negative priming tasks 22 Footnotes

1

However, to be honest, there are critical differences between these experiments. In

the experiments of Milliken et al. (1998) a target stimulus was never presented in the prime display whereas in the experiments of Allport et al. (1985) a target stimulus was always presented in the prime display. Additionally, the time interval between prime and probe display varied between both studies (about 500 ms compared to 1200 ms). Most importantly, participants in the experiments of Allport et al. tried to identify the stimuli of the prime display, that is they knew that there were stimuli presented whereas the participants of Milliken and colleagues did not know or were not given any hint to assume that masked stimuli were presented. It may be that these differences have caused the different priming effects.
2

A is the non-parametric signal detection sensitivity index that is typically used if the number

of observations is very small or if the hit rates of some participants are perfect. Note that chance performance yielded an A of 0.5, whereas perfect performance was reflected in an A value of 1.0.

Milliken et al. (1998) as well as Neill and Kahan (1999) mentioned the possibility that

participants inhibited the complete prime display (because the prime display did not match the internally defined template of a target stimuli, that is a clear readable, red words). If one adopts this perspective, NP with masked distractor-only prime trials can be explained with a more general inhibition account.

awareness in negative priming tasks 23 Table 1 Mean Response Times for the Repeated and Unrepeated Prime Conditions as a Function of Presentation Mode (Experiment 1) Presentation Subliminal Repeated Unrepeated NP effecta 815 808 -7*b (4) Supraliminal 810 801 -9* (4) 813 806 -8** (3) overall

* p < .05 ** p < .01 a RT(Unrepeated) minus RT(Repeated); Standard errors in Parentheses b one-tailed test

awareness in negative priming tasks 24 Table 2 Mean Response Times for the repeated and unrepeated prime conditions as a function of prime-probe color match (Experiment 2)

Unaware (N = 16) Prime-Probe-match Same color Repeated Unrepeated NP effecta 642 629 -14* (5) Different color 644 633 -11*b (6) 643 631 -12* (5) overall

Aware( N = 4) overall 635 651 16* (3)

* p < .05 a RT(Unrepeated) minus RT(Repeated); Standard errors in Parentheses b one-tailed test