Negative Masked Semantic Priming Center-Surround or ... - CiteSeerX

Repeated Masked Semantic Priming

1

RUNNING HEAD: Negative Masked Semantic Priming

Center-Surround or Spreading Inhibition: Which Mechanism Caused The Negative Effect From Repeated Masked Semantic Primes?

Christian Frings, Christina Bermeitinger, and Dirk Wentura

Saarland University, Saarbrücken, Germany

Word count (main text) 4672 Correspondence address: Christian Frings Saarland University Departement of Psychology Building A2 4 P.O. Box 15 11 50 D – 66041 Saarbrücken Germany Fon: 0049 681 302 6577 Fax: 0049 681 3024049 Mail: [email protected]


2 Abstract

In the paradigm of repeated masked semantic priming (Wentura & Frings, 2005, JEP:LMC) prime and mask are repeatedly and rapidly interchanged. Using this technique in a semantic priming task with category labels as primes and category exemplars as targets (related, e.g., BIRD – Swan, unrelated, e.g., BIRD - Lily), Wentura & Frings found a negatively signed priming effect. Here we used the repeated masking technique with category exemplars as targets and primes (i.e., identity priming) for analyzing, whether this effect reflects center-surround or spreading inhibition. If the repeated masked technique reflects spreading inhibition, a negative effect should also appear for identity priming. In contrast, a center-surround approach would predict a positive effect. In accordance with the latter hypothesis, we found a significant positive effect in identity priming (Experiment 1a) and significant difference to the negatively signed semantic priming effect when primes were category labels (Experiment 1b). This is indicative of the repeated masked semantic priming effect being a negatively signed semantic priming effect due to a center-surround mechanism.

Key words: semantic priming, repeated masked semantic priming, negative priming, center surround inhibition, category priming

Repeated Masked Semantic Priming Center-Surround or Spreading Inhibition: Which Mechanism Caused The Negative Effect From Repeated Masked Semantic Primes? Semantic Priming constitutes a classical issue in cognitive psychology (for reviews see Neely, 1991; Lucas, 2000; McNamara & Holbrook, 2003; McNamara, 2005). Usually, a prime stimulus (e.g., a word) is presented before a target stimulus (e.g., another word) while participants have to react only to the target (e.g., categorize it as a word or nonword). Typically, reaction times are facilitated, if both stimuli are in an associative or semantic relation (e.g., FRUIT – apple) compared to unrelated prime-target pairs. This phenomenon has been widely analyzed since most researchers agree that the semantic priming paradigm can help to understand how semantic knowledge is represented in the human cognitive system. Recently, Wentura and Frings (2005) introduced a new variant of semantic priming that potentially opens up a new route to that issue. In exploring the possibility of masked semantic priming (see, e.g., Klinger, Burton, & Pitts, 2000; Bodner & Masson, 2003), they observed a negatively signed semantic priming effect (i.e., slower responses to related than unrelated targets) for low prototypical (i.e., low dominant) category exemplars if primed by the category name (e.g., FRUIT – mango). For high prototypical (i.e., high dominant) exemplars, across experiments, either a null effect or a positive effect was found. This effect was observed with a new technique to present the primes, that is, by interchanging prime and mask rapidly and repeatedly. Thus, the summed prime duration of the masked prime was hold as long as in unmasked studies albeit most participants could not discriminate the primes above chance. Although Wentura and Frings (2005) suggested a center-surround inhibition mechanism for explaining their results (see below for a detailed description) the

3

Repeated Masked Semantic Priming observed data pattern can also be explained in terms of an alternative inhibition process, namely the spreading inhibition mechanism. This inhibition mechanism is usually analyzed in selective attention research, in the so called semantic Negative Priming (NP) paradigm were the effects from ignored primes on subsequent semantically related targets are of interest (Tipper 1985; Houghton & Tipper, 1994; Tipper, 2001). Several research groups argued (e.g., Yee, 1991; Fuentes & Tudela, 1992; Fox, 1994, 1996; Marí-Beffa, Fuentes, Catena, & Houghton, 2000; Hutchison, 2002; Ortells, Abad, Noguera, & Lupiáñez, 2001), that ignoring a prime word leads to an active inhibition of the representation of this stimulus and that inhibition spreads through the semantic network to related concepts like spreading activation does (e.g., Anderson, 1983). Access to inhibited representations is hampered and hence reactions to these stimuli are slowed. For example, inhibiting the word CAT leads to slower reactions to the word DOG in the following trial due to spreading inhibition (cf., Hutchison, 2002).1 Then, however, the results of Wentura and Frings (2005) could also be explained by spreading inhibition. Of course, one must assume that participants have ignored the masked category prime. Yet, this is very plausible, since participants were told that they should attend only to the target. The brief flicker, (i.e., the mask and the prime presented before the target) was not explicitly mentioned and hence most participants would probably have ignored it (see also Milliken et al., 1998, for the same argument). Hence a spreading inhibition account would state that the prime word with the Wentura and Frings (2005) technique became inhibited and that related concepts also became inhibited due to spreading inhibition. That is, when participants ignored the masked category word FRUIT, the representation of this stimulus and all category coordinates should

4

Repeated Masked Semantic Priming become inhibited and reactions to these stimuli following immediately after prime presentation should be slowed. Thus, all experiments reported in the Wentura and Frings (2005) study could be explained in terms of a spreading inhibition account as well. It must be noted, that the inhibition account from the NP paradigm usually hinges on the presence of distractors which are presented simultaneously with the target (e.g., Moore, 1994) which was not the case in the experiments by Wentura and Frings (2005). There are conditions, however, under which this does not hold true. In particular, if the ignored word and the following target had a semantic relationship as in the Wentura and Frings (2005) study, inhibition effects were observed in experiments in which targets were never accompanied by distractor stimuli (e.g., Yee, 1991; Ortells et al., 2001). Moreover, Frings and Wentura (2006) have argued that the dependence on the presence of distractors is modulated by strategies based on the ignored prime-target relationship (see also Christie & Klein, 2001). In a nutshell, with semantically related ignored primetarget pairs the chances to observe an inhibition effect without distractors are higher than with identically related pairs. In sum, the negative priming effects from masked repeated semantic primes could therefore be explained by spreading inhibition although target stimuli were never accompanied by distractors. In contrast, Wentura and Frings interpreted this effect as evidence for the center-surround inhibition theory of Dagenbach and colleagues (e.g., Carr & Dagenbach, 1990; Dagenbach, Carr, & Barnhardt, 1990; but see Kahan, 2000). This theory assumes that a weakly activated node (e.g., resulting from a masked stimulus) is surrounded by a ‘ring of inhibition’; due to the activation differences between the weakly active stimulus (the prime) and the inhibited surrounding, the prime can eventually become accessible; after successful access, inhibition is

5

Repeated Masked Semantic Priming dismissed. Thus, masked category primes which were not clearly accessible should be surrounded by inhibition and hence presenting category coordinates from this category (which possibly lay in this surround) as the following target should result in RT costs. In contrast, masked primes that clearly activate their corresponding code should not lead to cost effects since then the center-surround mechanism is not needed and hence no inhibition occurs. Indeed, Wentura and Frings (2005) observed a data pattern compatible with this prediction: participants with above random responding in a subsequent direct test of prime awareness showed no priming effects (or even small positive effects) whereas participants with random responding showed negative priming effects. Probably, for the first subsample the activation caused by the masked prime was strong enough to make the center-surround inhibition unnecessary. Which inhibition mechanism has caused the negative effects from repeated masked primes observed in Wentura and Frings (2005)? Center-surround inhibition is a mechanism assumed to be applied to the structure of semantic memory for gaining access to weakly activated nodes. The negative priming effect from repeated masked primes would therefore reveal something about the representation of categories and the processes that operate on these representations. In contrast, spreading inhibition is assumed to result from ignoring a stimulus and taps processes of selective attention. The aim of the present study is therefore to decide whether center-surround inhibition or spreading inhibition is suited to explain repeated masked semantic priming. As discussed, both theories can explain negative effects from categorically related prime-target pairs. However, they differ completely in what they predict for identically related prime-target pairs. In this case, center-surround theory predicts null or even positive results since the center itself should never be

6


7

inhibited (Dagenbach et al., 1989; Kahan, 2000) whereas spreading inhibition predicts even stronger negative effects from the initially inhibited word than from related concepts (Fox, 1995; May et al., 1995). Thus, in Experiment 1a primes and targets were identically related. Yet, for an even more conservative test, we additionally presented distractors in the target display. As discussed above, although there is evidence that effects of spreading inhibition can be obtained without probe distractors (e.g., Ortells et al., 2001; Yee, 1991; see Frings & Wentura, 2006), there is also consensus that adding these distractors to the target display makes the effect more reliable. A null effect in an identity priming design without probe distractors would be open to severe critic because by far most (identity) negative priming results were found in the presence of probe distractors. Thus, we changed the procedure of Wentura and Frings (2005) in such a way that the ‘optimal’ conditions for observing spreading inhibition were used. To relate Experiment 1a even more closely to the studies by Wentura and Frings (2005), we replicated Wentura and Frings in Experiment 1b by using primes and targets that were categorically related and by using exactly the same experimental program as in Experiment 1a. In accordance with Wentura and Frings and the tradition of semantic priming research, targets were not accompanied by distractors. Method Participants In Experiment 1a, thirty-one students (27 women) from Saarland University which were naive to the purpose of the study participated in the experiment. The median age was 22 years (ranging from 19 to 33 years). In Experiment 1b, thirty naïve students (16 women) from Saarland University participated in the experiment. The median age was 22 years (ranging from 19 to 35 years). All

Repeated Masked Semantic Priming participants were native speakers of German and had normal or corrected-tonormal vision. They received 5 € for participation. Design Essentially, the experiments rested on a three-factorial design. The first factor was priming condition (unrelated, related, and neutral prime-target pairs). The second factor was dominance of the target exemplars (high dominant versus low dominant exemplar of its category). The third factor was target-orthography (word versus non-word). All factors were varied within subjects. Analyses were focused on word trials; yet, non-word trials are reported in the Appendix. Priming effects were computed as the difference between related and unrelated primetarget pairs. Material The same material as in Wentura and Frings (2005) was used in both experiments. That is, we used exemplars from four categories (birds, insects, flowers, and fruits). Three high dominant and three low dominant exemplars of each category served as targets. High dominant exemplars had a mean association frequency of 67.1 % (SD = 10.7 %; range 55 % to 86.5 %), whereas low dominant exemplars had a mean association frequency of 6.2 % (SD = 2.87 %; range 2.5 % to 11.5 %; Mannhaupt, 1983). Mean length was 5.2 (SD = 0.8; range 4 to 7) for the high dominant exemplars and 5.4 (SD = 0.5; range 5 and 6) for the low dominant exemplars. The average of word frequency was 5318 (SD = 10026) for high dominant exemplars and 502 (SD = 727) for low dominant exemplars (according to the German database of written language, COSMAS II). Pronounceable non-words were created by changing one letter of each target word.

8

Repeated Masked Semantic Priming In Experiment 1a, the target words served as primes as well (see Procedure for details). The distractors flanking the targets were chosen from four new categories (trees, furniture, clothes, and instruments). Three high dominant and three low dominant distractors were chosen from each category. Distractors did not significantly differ from targets with respect to association frequency, word frequency, and word lengths (all ps > .26). Distractor words had exactly the same font properties and size as targets. A pronounceable non-word was formed for each distractor word by changing one letter of the word. As in Wentura and Frings (2005), the category names (i.e., BIRD, INSECT, FLOWER, and FRUIT) served as primes in Experiment 1b (see Procedure for details.) All prime stimuli were presented in light grey (about 41 cd/m²) on black screen and were approximately 0.5 cm (0.48°) in height. Primes and masks were approximately 3.0 cm (2.86°) in width; targets were between 1.4 cm (1.34°) and 2.6 cm (2.48°) in width. Targets in Experiment 1a were red and flanked by green distractors. Targets in Experiment 1b were presented in light grey. Procedure Participants were tested individually in sound-attenuated chambers. The experiment was run using the E-Prime software (version 1.1) with a standard PC and 17’’ CRT monitors. Instructions were given on the CRT screen from which the participants had a distance of approximately 60 cm. Participants were told that target-words belonging to the categories birds, insects, flowers, and fruits would be presented on the screen. Some of these words, however, would be written with a spelling mistake. They were requested to quickly and correctly categorize each word with regard to orthography by pressing either the right key with their right index finger for correctly written words or the left key with their left index finger for misspelled words. Following Wentura and Frings (2005), the sequence of each

9

Repeated Masked Semantic Priming trial was as follows (see Figure 1): First a fixation stimulus (+) appeared at the screen-center for 500 ms. It was followed by a forward mask, consisting of eight randomly generated capital consonants, which was presented for one refresh cycle (i.e., about 13 ms). The forward mask was immediately overwritten by the prime which was on the screen for the next refresh cycle. Primes were written in capital letters (Font: Fixedsys). In Experiment 1a, the related prime was always identical to the target. The unrelated prime was always an insect exemplar for fruit targets, a fruit exemplar for insect targets, a bird exemplar for flower targets, and a flower exemplar for bird targets. Unrelated primes and targets were matched for dominance. In Experiment 1b, the related prime was always the category name that corresponded to the target. The unrelated prime was always INSEKT (insect) for fruit exemplars, FRUCHT (fruit) for insect exemplars, VOGEL (bird) for flower exemplars, and BLUME (flower) for bird exemplars. Neutral primes were randomly created strings of five letters2. Random capital consonants were added to the left and to the right of the prime to create a string of eight letters3. The prime was followed by a backward mask consisting of eight random consonants for one refresh cycle. Prime and backward mask were alternately presented for a total of 20 refresh cycles. Thus, the total prime duration was 133 ms. After the last backward mask, the target appeared and remained on the screen until a response was given. In the case of a false response, an error message was given on the screen until a keypress. The intertrial interval with a blank screen was 1000 ms. In Experiment 1a, we added distractors to the target display (see Introduction). These distractors were chosen from four new categories (trees, furniture, clothes, and instruments). Each target was presented twice (once with a distractor word and once with a distractor non-word) within each priming condition (related, unrelated, neutral). Within one trial, distractor and target

10


11

always had the same dominance. On each trial, distractors appeared directly above and below the target (see Figure 1). The experiment comprised two blocks with 144 trials each (48 related, 48 unrelated, and 48 neutral prime-target pairs). Within a block, each exemplar stimulus was presented thrice as a target. The sequence of priming conditions for a given target was determined by a Latin-square design (i.e., sequence of targets and conditions was balanced over participants). Throughout the experiment, each stimulus served four times as a related prime and four times as the unrelated prime (once for each combination of word/non-word status of target and distractor). Participants could make a rest after every 24 trials. Before the experimental trials, participants practiced the task with 48 practice trials. In accordance with Wentura & Frings (2005), Experiment 1b comprised three blocks with 48 trials each (16 related, 16 unrelated, and 16 neutral primetarget pairs) because it did not need the manipulation of the word/non-word status of the distractor. Everything else was essentially the same for both experiments. After the priming task, a direct test of prime categorization was administered. In Experiment 1a, 48 more trials (two trials for each prime word) were presented in the same manner like the experimental trials, but with a row of question marks instead of a target. The prime and the corresponding exemplar which had served as the unrelated prime for that stimulus (see above) were presented (with balanced allocation) to the left and right of the question marks, respectively. Participants had to categorize the prime by pressing the corresponding key. In Experiment 1b, thirty-two trials (eight trials for each prime word) were presented accordingly. The direct test was practiced with 10 trials. Results

Repeated Masked Semantic Priming Unless otherwise noted, all effects referred to as statistically significant throughout the text are associated with p-values of less than .05, two-tailed. Direct test In line with Wentura and Frings (2005), we looked at the individual contingencies between prime category and response category, that is, we analyzed for each participant whether s/he showed an above chance categorization of masked primes. Participants with an above chance discrimination rate were labelled ‘high-D’ (high discrimination) participants whereas participants with a chance discrimination rate were labelled ‘low-D’ (low discrimination) participants. In Experiment 1a, seven participants showed a significant contingency between prime and response, all χ2> 4.5, p < .03 and were hereby classified as high-D. The remainder of the sample (n = 24) showed no significant contingency between prime and response; all individual χ2 < 3.24, p > .07. These participants were classified as low-D. For both samples we also computed the parametric signal detection sensitivity index d’ with hits being correctly identified words and false alarms being incorrectly identified words (Green & Swets, 1966); the high-D group had an average d’ of 1.38 (SD = 0.26, t[6] = 14.30, p < .001) and the low-D group an average d’ of 0.30 (SD = 0.38, t[23] = 3.83, p < .01). In Experiment 1b, five participants showed a significant contingency between prime and response, all χ2 > 4.5, p < .03. These participants were thereby classified as high-D participants. The remainder of the sample (n = 25) showed no significant contingency between prime and response, all individual χ2 < 3.24, p > .07, and was thereby classified as low-D participants. The high-D group had an average d’ of 1.29 (SD = 0.33, t[4] = 8.79, p < .01) and the low-D group an average d’ of 0.26 (SD = 0.33, t[24] = 3.83, p < .01). In accordance to Wentura and Frings

12

Repeated Masked Semantic Priming (2005), our analyses focussed on participants who had no explicit access to the lexicographic status of the prime (i.e., low-D participants); however, for the sake of completeness we also report data from high-D participants. Priming effects Mean RTs were derived from correct responses to word trials. The mean error rates were 2.89 % (Experiment 1a) and 3.33 % (Experiment 1b). RTs that were 1.5 interquartile ranges above the third quartile with respect to the individual distribution (Tukey, 1977) or were below 200 ms were discarded. RTs higher than 1500 ms were generally discarded. These criteria led overall to the exclusion of 7.33 % (Experiment 1a) and 8.89 % (Experiment 1b) trials. Mean RTs, and mean error rates for word targets are shown in Table 1. Mean RTs and mean error rates for non-word trials for both experiments are presented in the Appendix. Experiment 1a. Mean RTs of low-D participants were submitted to a 2 (priming condition: identical versus unrelated) x 2 (target dominance: high versus low) ANOVA.4 A significant main effect for priming condition emerged, F(1,23) = 4.31, p < .05. In accordance with the center-surround hypothesis, but in contrast to the spreading inhibition hypothesis, this effect indicated a significant positive instead of a negative priming effect of M = 9 ms (SD = 22 ms), that is participants responded slower to unrelated than to repeated targets. A significant main effect for dominance emerged, F(1,23) = 47.96, p .76. The same ANOVA on mean RTs and error rates from high-D participants revealed a significant main effect only for target dominance in the analyses of mean RTs, F(1,6) = 25.17, p < .01, all other ps > .06.

13

Repeated Masked Semantic Priming Experiment 1b. Mean RTs of low-D participants were submitted to a 2 (priming condition: related versus unrelated) x 2 (target dominance: high versus low) ANOVA. A significant main effect for priming condition emerged, F(1, 24) = 5.17, p < .05. Replicating Wentura and Frings (2005), participants responded on average M = 16 ms (SD = 34 ms) slower to related targets than to unrelated targets. The main effect for dominance was also significant, F(1,24) = 33.30, p < .001, indicating that participants reacted in general quicker to high dominant (673 ms) than low dominant (714 ms) targets. Yet, priming condition and dominance did not interact, F(1,24) = 0.04, p = .84. Analyses on error rates revealed only a main effect for target dominance, F(1,24) = 12.44, p < .02, both other ps > .38. The same analysis on mean RTs and error rates from high-D participants revealed only a significant main effect for target dominance in the analyses of mean RTs, F(1,4) = 17.57, p < 05, all other ps > .09. Combined Analysis of Experiments 1a and 1b. The main interest was in the comparison of priming effects in dependence of prime-target relationship. Yet, given that dominance did not moderate priming in any experiment, we dropped this factor for the overall-analysis. Thus, we subjected the mean reaction times of low-D participants from Experiment 1a and from Experiment 1b to a 2 (priming condition: related vs. unrelated) x 2 (prime-target relation: identity vs. semantic) mixed ANOVA. The main effect for priming condition was not significant, F(1,47) = 0.59, p = .45 as well as the main effect for prime-target relation, F(1,47) = 1.86, p = .18. However, there was a significant interaction of priming condition and prime-target relation, F(1,47) = 9.07, p < .01. This interaction reflected the fact that we observed significant positive priming effects when primes and target were identical (Experiment 1a) but negative priming effects when primes and

14

Repeated Masked Semantic Priming targets were semantically related (Experiment 1b). The same analysis on error rates yielded no significant effects, all Fs < 1.64, all ps > .21. Discussion With the presented studies, we focussed on the question which inhibition mechanism has caused the negative effects from repeated masked primes observed in Wentura and Frings (2005). Center-surround inhibition, as suggested by the authors, is a within-semantic memory mechanism that could possibly shed some light on the representation of categories and the processes that operate on these representations. However, from a negative priming perspective, one could argue that the effect results from ignoring a stimulus and the subsequent process of spreading inihibition. Therefore, this effect would indicate processes of selective attention. With Experiment 1a, we tested the two competing hypotheses. Related primes were now identical to the targets. The hypothesis of spreading inhibition predicts a negative priming effect, whereas positive effects (or null effects) are expected from the center-surround inhibition hypothesis. In fact, we observed a significantly positive effect. It is important to note, that we modelled Experiment 1a in such a way that optimal conditions for observing negative effects from spreading inhibition were used: We presented targets accompanied by distractors. In the field of negative priming research, the presence of distractors is seen as a clear means to boost (negative) effects from spreading inhibition. Note, however, that we also ran a further control experiment in which we never presented distractors but still used an identity relation with the same procedure and material as in Experiment 1a. In this experiment, we obtained a positive trend instead of a negative effect for identically related prime-target pairs.

15

Repeated Masked Semantic Priming In addition, the comparison of Experiment 1a and 1b revealed clear evidence for a center-surround inhibition mechanism. As stated in the introduction, from a center-surround inhibition we expected an interaction of priming condition and prime-target relation (indicating negative effects with categorically related but positive or null effects with identically related primetarget pairs) whereas from spreading inhibition we should have found only a main effect for priming condition (indicating negative effects irrespective of primetarget relation). In fact, we observed a significant interaction but no significant main effect for priming condition; thus, there was a clear dissociation between semantically related (Experiment 1b) and identity related prime-target pairs (Experiment 1a) with the repeated-mask technique which further confirms that the negative repeated masked semantic priming effect is no instance of spreading inhibition. Instead, the negatively signed semantic priming effect found by Wentura and Frings (2005) and in the present Experiment 1b could be explained of a center-surround mechanism which stems from the tradition of semantic priming. The results of Experiment 1b replicate the main finding from Wentura and Frings (2005): there was a negative semantic priming effect from repeated masked primes. Yet, in slight deviance to Wentura and Frings (2005) who found the negatively signed effect most markedly for low dominant targets, Experiment 1b revealed a negative effect that was not moderated by dominance. Participants who responded at chance level in a direct test showed increased lexical decision responses to category coordinates preceded by their category labels independently of targets’ category dominance. As mentioned in Wentura and Frings (2005), regarding the center-surround theory as a possible explanation for negative effects from repeated masked semantic priming, there is some arbitrariness in the

16

Repeated Masked Semantic Priming assumption that high dominant exemplars are in the center (and thus never inhibited) whereas low dominant exemplars are in the surround (and thus inhibited) of their category. To reconcile the results of Experiment 1b with the center-surround inhibition theory, we have to assume interindividual differences in the representation of categories; that is, we assume that for some participants the surround might comprise all exemplars of a category and the center is only the primed concept itself whereas for other participants high dominant category coordinates belong still to the center of a category. Yet, it must be acknowledged that this is a post hoc explanation and thus future research should analyze conditions under which high dominant targets are negatively or positively primed by masked repeated category primes. However, the main finding is the contrast between positive identity priming (Experiment 1a) and negative semantic priming (Experiment 1b). There is a long going debate in cognitive psychology whether behavior can be influenced by stimuli that are presented subliminally. Especially, the criteria for ‘truly subliminal perception’ were discussed (e.g., Holender, 1986; Merikle, Smilek, & Eastwood, 2001; Forster, Mohan, & Hector, 2003). With respect to this debate, repeated masked priming would clearly not be considered as a truly subliminal presentation. For example, mean d’ of the overall sample in the experiments of Wentura and Frings (2005) and in the present experiments was clearly above zero. Thus, we label primes presented with this technique ‘marginally perceptible’ or ‘near threshold’. Note, however, that for our effect this debate is of no relevance. Over experiments, we found a qualitative difference in repeated masked priming in dependence of participants’ ability to discriminate primes in a direct test after the experiments; that is participants with a low discrimination performance showed this negative effect whereas participants with

17

Repeated Masked Semantic Priming a high discrimination performance showed positive or null results. The interesting point is that the special kind of prime presentation turns the supraliminal positive priming effect into a negative one for participants with a low discrimination performance. It must be acknowledged, however, that it remains open for future research to identify possible interindividual differences – beyond differences in the performance of the direct test – which contribute to this dissociation. Furthermore, in recent years several studies replicated positive effects from masked semantic primes (e.g. Kiefer, 2002; Kiefer & Spitzer, 2000; Kiefer & Brendel, 2006), as originally observed by Marcel (1983). In fact, Wentura and Frings (2005; Exp. 1) dissociated masked repeated semantic priming from masked semantic priming. They observed negative effects from repeated masked primes but a positive (albeit not significant) effect from masked primes. Thus, repeated masked priming taps probably another process as a standard masked priming procedure. With respect to the suggested center-surround mechanism it could be argued, that this mechanism is only activated when marginally perceptible information is encountered. For gaining a faster access to the weak activated stimulus representation the surround becomes inhibited. In the case of standard masked priming (with subliminal primes), however, unconscious activation of related concepts is triggered. Overall, we demonstrated here that negatively signed priming effects from repeated masked primes are a replicable and new phenomenon that cannot be explained by an alternative spreading inhibition account that stems from the field of negative priming research. So far, the center-surround hypothesis is the most plausible approach to the understanding of this effect.

18


19

References Anderson, J.R. (1983). The architecture of cognition. Boston: Harvard University Press. Bodner, G. E., & Masson, M. E. J. (2003). Beyond spreading activation: An influence of relatedness proportion on masked semantic priming. Psychonomic Bulletin & Review, 10, 645-652. Carr, T. H., & Dagenbach, D. (1990). Semantic priming and repetition priming from masked words: Evidence for a center-surround attentional mechanism in perceptual recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 341-350. Christie, J. & Klein, R.M. (2001). Negative priming for spatial location ? Canadian journal of Experimental Psychology, 55(1), 24-38. Dagenbach, D., Carr, T. H., & Barnhardt, T. M. (1990). Inhibitory semantic priming of lexical decisions due to failure to retrieve weakly activated codes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 328-340. Dagenbach, D., Carr, T. H., & Wilhelmsen, A. (1989). Task-induced strategies and near-threshold priming: Conscious influences on unconscious perception. Journal of Memory and Language, 28, 412-443. Forster, K., Mohan, K., & Hector, J. (2003). The mechanics of masked priming. In S. Kinoshita & S. Lupker (eds.), Masked priming: State of the art. Hove: Psychology Press. Fox, E. (1994). Attentional bias in anxiety: A defective inhibition hypothesis. Cognition & Emotion, 8, 165-195.

Repeated Masked Semantic Priming Fox, E. (1995). Negative priming from ignored distractors in visual selection: A review. Psychonomic Bulletin & Review, 2, 145-173. Fox, E. (1996). Cross-language priming from ignored words: Evidence for a common representational system in bilinguals. Journal of Memory and Language, 35, 353-370. Frings, C. & Wentura, D. (2006). Strategy effects counteract distractor inhibition: Negative priming with constantly absent probe distractors. Journal of Experimental Psychology: Human Perception and Performance, 32, 854-864. Fuentes, L. J., & Tudela, P. (1992). Semantic processing of foveally and parafoveally presented words in a lexical decision task. The Quarterly Journal of Experimental Psycholgy, 45(A), 299-322. Holender, D. (1986). Semantic activation without conscious identification in dichotic listening, parafoveal vision, and visual masking: A survey and appraisal. Behavioral and Brain Sciences, 9, 1-66. Houghton, G., & Tipper, S. P. (1994). A model of inhibitory mechanisms in selective attention. In D. Dagenbach & T. H. Carr (Eds.), Inhibitory processes in Attention, Memory, and Language (pp. 53-112). San Diego, CA, US: Academic Press. Hutchison, K. A. (2002). The effect of asymmetrical association on positive and negative semantic priming. Memory & Cognition, 30, 1263-1276. Jonides, J. & Mack, R. (1984). On the cost and benefit of cost and benefit. Psychological Bulletin, 96, 29-44. Kahan, T. A. (2000). Negative priming from masked words: Retrospective prime clarification or center-surround inhibition? Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 1392-1410.

20

Repeated Masked Semantic Priming Kiefer, M. (2002). The N400 is modulated by unconsciously perceived masked words: Further evidence for an automatic spreading activation account of N400 priming effects. Cognitive Brain Research, 13, 27-39. Kiefer, M. & Brendel, D. (2006). Attentional modulation of unconscious ‘automatic’ processes: Evidence from event-related potentials in a masked priming paradigm. Journal of Cognitive Neuroscience, 18, 184-198. Kiefer, M. & Spitzer, M. (2000). Time course of conscious and unconscious semantic brain activation. NeuroReport, 11(11), 2401-2407. Klinger, M. R., Burton, P. C., & Pitts, G. S. (2000). Mechanisms of unconscious priming: I. Response competition, not spreading activation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 441-455. Lucas, M. (2000). Semantic priming without association: A meta-analytic review. Psychonomic Bulletin & Review, 7, 618-630. Marcel, A.J. (1983). Conscious and unconscious perception: experiments on visual masking and word recognition. Cognitive Psychology, 15(2), 197-237. Marí-Beffa, P., Fuentes, L. J., Catena, A., & Houghton, G. (2000). Semantic priming in the prime task effect: Evidence of automatic semantic processing of distractors. Memory & Cognition, 28, 635-647. May, C. P., Kane, M. J., & Hasher, L. (1995). Determinants of negative priming. Psychological Bulletin, 118, 35-54. McNamara, T. P. (2005). Semantic priming: Perspectives from memory and word recognition. New York: Psychology Press. McNamara, T. P., & Holbrook, J. B. (2003). Semantic memory and priming. In I. B. Weiner (Series Ed.), A. F. Healy, & R. W. Proctor (Vol. Eds.), Handbook of psychology: Experimental psychology (pp. 447-474). New York: Wiley & Sons.

21

Repeated Masked Semantic Priming Merikle, P. M., Smilek, D., & Eastwood, J. D. (2001). Perception without awareness: Perspectives from cognitive psychology. Cognition, 79, 115-134. Milliken, B., Joordens, S., Merikle, P. M., & Seiffert, A. E. (1998). Selective attention: A reevaluation of the implications of negative priming. Psychological Review, 105, 203-229. Moore, C. M. (1994). Negative priming depends on probe-trial conflict: Where has all the inhibition gone? Perception & Psychophysics, 56, 133-147. 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. Ortells, J. J., Abad, M. J. F., Noguera, C., & Lupiáñez, J. (2001). Influence of prime-probe stimulus onset asynchrony and prime precuing manipulations on semantic priming effects with words in a lexical-decision task. Journal of Experimental Psychology: Human Perception and Performance, 27, 75-91. .Tipper, S. P. (1985). The negative priming effect: Inhibitory priming by ignored objects. The Quarterly Journal of Experimental Psychology, 37(A), 571-590. Tipper, S. P. (2001). Does negative priming reflect inhibitory mechanisms? A review and integration of conflicting views. The Quarterly Journal of Experimental Psychology, 54(A), 321-343. Tukey, J. W. (1977). Exploratory data analysis (pp. 1-56). Reading, MA: Addison-Wesley. Wentura, D., & Frings, C. (2005). Repeated masked category primes interfere with related exemplars: New evidence for negative semantic priming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 108-120.

22

Repeated Masked Semantic Priming Yee, P. L. (1991). Semantic inhibition of ignored words during a figure classification task. The Quarterly Journal of Experimental Psychology, 43(A), 127-153.

23

Repeated Masked Semantic Priming Author note Christian Frings, Christina Bermeitinger, and Dirk Wentura, Department of Psychology, Saarland University, Saarbrücken, Germany. The research reported in this article was supported by a grant of the Deutsche Forschungsgemeinschaft to Dirk Wentura and Christian Frings (WE 2284/ 5-2). Correspondence concerning this article should be addressed to Christian Frings, Saarland University, Department of Psychology, Building A2 4, P.O. Box 15 11 50, D – 66041 Saarbrücken, Germany or via email to [email protected].

24

Repeated Masked Semantic Priming Footnotes 1

We should add that most of the negative priming research is concerned with

identity prime-target relations. Whereas for this part of NP research there is an ongoing lively debate about whether NP effects should be better explained by inhibition or memory retrieval processes, semantic NP is typically attributed to processes of spreading inhibition. 2

We included neutral primes to mimic the former experiments as closely as

possible. However, we do not want to put emphasis on these trials because the results for neutral conditions are always somewhat ambiguous (e.g., Jonides & Mack, 1984). 3

Due to the random flanking letters, in 7-10% of all trials additional prime words

were created. Thus, in these trials it remains somewhat unclear which prime words were actually primed. 4

The orthography of the distractors did not interact with any of the other factors

and thus this factor was omitted.

25


26

Table 1 Mean Response Times (in ms) and Error rates (in percentage) of Word Trials as a Function of Prime Condition, Category Dominance of Target Exemplars, and Participants’ Prime Discrimination Performance in, Experiments 1a & 1b. Low-D Participants

High-D Participants

Related Unrelated Neutral


Experiment 1a Low dominant High dominant

674 (4.0) 624 (0.7)

688 (4.0) 629 (1.0)

680 (3.5) 646 (0.9)

622 (5.4) 562 (2.4)

615 (7.7) 571 (1.8)

606 (6.6) 561 (4.8)

Experiment 1b Low dominant

723 705 722 766 781 762 (6.0) (6.3) (4.7) (1.7) (3.3) (6.7) High dominant 680 666 673 715 747 730 (1.3) (0.3) (2.3) (0.0) (1.7) (1.7) Note. The number of high-D participants were 7 out of 31 in Experiment 1a, and 5 out of 30 in Experiment 1b.

27

Negative Masked Semantic Priming

Figure captions.

Figure 1. Schematically order of an experimental trial in Experiment 1a and Experiment 1b. The example shows a related trial for the category INSEKT (insect) with a high dominant target word FLIEGE (fly). See procedure sections for more details.

Figure 2. Priming effects in Milliseconds for low-D participants as a function of prime-target relationship, identically (Experiment 1a) and categorically related (Experiment 1b).


28 Figure 1

Experiment 1a

Experiment 1b

+

+

MSZPTRMK

MSZPTRMK

Fixation cross, 500 ms Forward mask, 13 ms

ZFRUCHTK

ZBANANEK Prime, 13 ms

VKTPWMBL

VKTPWMBL Backward mask, 13 ms

ZBANANEK

Prime, 13 ms

ZFRUCHTK

VKTPWMBL

VKTPWMBL

Backward mask, 13 ms Seven more times alternations of prime and backward mask ZBANANEK Prime, 13 ms

ZFRUCHTK

VKTPWMBL

Backward mask, 13 ms Target, until response

Sessel Banane Sessel

VKTPWMBL Banane

29

Figure 2

20 15

Priming Effect in ms

10 5 0 -5 -10 -15 -20 -25 -30

Experiment 1a

Experiment 1b



30

Appendix A Mean Response Times (in ms) and Error rates (in percentage) of Non-word Trials as a Function of Prime Condition, Category Dominance of Target Exemplars, and Participants’ Prime Discrimination Performance in Experiments 1a & 1b. Low-D Participants

High-D Participants



Experiment 1a Low dominant High dominant

727 (3.0) 708 (3.1)

725 (2.6) 702 (3.3)

724 (5.4) 701 (3.5)

633 (5.4) 627 (5.4)

645 (5.4) 635 (2.4)

633 (3.0) 628 (0.6)

Experiment 1b Low dominant

758 745 768 778 829 819 (4.0) (4.0) (3.0) (3.3) (3.3) (1.7) High dominant 746 744 751 802 721 779 (2.7) (3.0) (2.3) (0.0) (1.7) (0.0) Note. The number of high-D participants were 7 out of 31 in Experiment 1a, and 5 out of 30 in Experiment 1b.