Inhibitory effects during object name retrieval: The effect of interval between prime and target on picture naming responses.
Three picture naming experiments are reported which examine the relationship between the apparent inhibition of a response on one trial, and naming latency on the subsequent trial. The design of each experiment involves the presentation of prime and target pairs, either presented in succession (Lag 1 condition), or separated by two intervening unrelated trials (Lag 3 condition). A control condition is also included. In Experiment 1, a speeded picture naming task is used, and naming errors are analysed. Target pictures are misnamed at above chance rates with the name of the semantically related prime picture in the Lag 3 condition. In contrast, these prime-related errors do not occur in the Lag 1 condition, suggesting a brief inhibitory effect. If primes are briefly inhibited, then target naming larencies immediately following a related prime should be quicker than target larencies in the Lag 3 condition. Experiment 2 confirms this pattern of results, using exactly the same stimuli and design, but standard nam ing instructions. Experiment 3 examines whether the inferred inhibition is the result of a self-inhibitory mechanism, using a repetition priming paradigm. If Lag 1 prime representations are self-inhibited, then facilitatory effects from prime/target repetition should be stronger in the Lag 3 condition, than in the Lag 1 condition. The data from Expt 3 were not consistent with this prediction. Taken together, the results of the three experiments suggest that a brief inhibitory effect occurs after retrieval of an object name, and that the inhibition may be accomplished by mechanisms other than self-inhibition.
A number of different studies have shown that object name retrieval can be impaired when the names of other objects from the same semantic category have earlier been retrieved (Brown, 1981; Kroll & Stewart, 1994; Vitkovirch & Humphreys, 1991; Wheeldon & Monsell, 1994). For example, Wheeldon and Monselt (1994) have shown that retrieving an object name from a definition can slow picture naming latencies when a semantically related picture is presented three or more trials later. Vitkovitch and Humphreys (1991) found that when participants named a block of target objects under speeded naming instructions, they frequently made errors which corresponded to semantically related primes which had been named in a previous block (perseverative errors). These results fit generally with the conclusion in the literature that object name retrieval is subject to semantic competition, shown by a number of other paradigms, such as picture--word interference tasks, (e.g. Glaser & Glaser, 1989; Srarreveldt & La Heij, 1995) or b y post-cue naming techniques (Humphreys, Lloyd-Jones, & Fias, 1995). The finding that an earlier named object interferes in particular with target name retrieval is thought to be due to the strengthening of an aspect of the prime's representation in memory, making it more likely to be sampled as a candidate response before the target name (Brown, 1981), or making it an especially strong competitor when it is activated, along with other competitors, in parallel with target representations (Humphreys, Riddoch, & Quinlan, 1988; Wheeldon & Monsell, 1994).
More recently, we have found that there is a distinct pattern apparent in the incidence of perseverarive errors. Vitkovitch, Kirby, and Tyrrell (1996) presented a sequence of related objects, and examined the perseverative naming errors as a function of lag between error and earlier response. While errors relating to responses at least four trials earlier (Lag 4) were evident at above chance rates, there was an absence of errors relating to the immediately preceding trial (Lag 1), and this was below chance rate. Very similar results have been found by Campbell and Clark (1989) in their analysis of errors made during the retrieval of answers to simple arithmetic problems (6 x 3, 7 x 9 etc). In line with Campbell and Clark (1989), we interpreted the negative error priming (i.e. absence of Lag 1 errors) as indicative of a brief inhibitory effect. Houghton & Tipper (1994) have recently referred to inhibition as a state of suppressed responsiveness, and it is in this sense that we use the term here. Once the inhi bitory effect is removed, the excitatory influences which cause the perseverative errors are revealed in above-chance error rates at longer lags (positive error priming).
There is increasing use of the metaphor of inhibition in the literature on cognition. For example, Tipper and Driver (1988) suggested that inhibitory processes can account for a negative priming effect found in a selective attention paradigm. Other areas of cognition which have implicated inhibitory processes are language production (e.g. Dell & O'Seaghda, 1994), episodic retrieval (Anderson & Bjork, 1994), and language comprehension (Gernsbacher & Faust, 1995). Houghton and Tipper (1996) have also recently suggested that inhibitory mechanisms may be useful in sequential tasks generally, to prevent reiteration of a previous response.
There are several mechanisms by which such inhibition might be achieved. Lateral inhibition from each node to others at a given level of representation has long been a feature of interactive activation models of stimulus recognition (McClelland & Rumelhart, 1981). Lateral inhibition may function to reduce the kind of interference that is caused by parallel activation of representations other than those of the target, as in the examples of semantic competition given above (Humphreys et al., 1988; Wheeldon & Monsell, 1994). Another possible means by which inhibition could be implemented is by representational nodes inhibiting themselves. This might be particularly useful in sequential tasks, and in language production in particular. Arbuthnott (1995) has presented a possible model of such self-inhibitory mechanisms. She also raises the issue of whether such inhibitory mechanisms are largely automatic or whether they can be modified as a result of strategic processing. A further possibility is that active suppr ession can operate in the absence of hard-wired inhibitory structures.
The three experiments reported here extend our previous work on inhibition during object name retrieval. The studies on naming errors (Vitkovitch & Humphreys, 1991; Vitkovitch et al., 1996) and naming latencies (e.g. Wheeldon & Monsell, 1994) seem to complement each other in showing semantic competition from earlier object naming trials. Campbell (1990) has examined the relationship between the apparent inhibition of representations on one arithmetic trial, and latencies to retrieve answers on the next trial. As yet, we have not examined latencies in relation to inhibition during object name retrieval. Vitkovitch et al. (1996), for example, looked at lag effects only in terms of the errors which emerged naturally as a result of speeded naming of a series of related pictures. The design precluded an analysis of response latencies in relation to lag. In the experiments below, we use a design similar to that employed by Wheeldon and Monsell (1994), and this enables us to manipulate the lag between related prime and target pictures and examine latencies. Experiment 1 replicates our previous work on detecting inhibition through analysis of error patterns using this different experimental design, and Expt 2 provides converging support for the presence of inhibition on prime trials through an analysis of naming latencies on target trials. Experiment 3 represents a preliminary investigation of the mechanism by which any such inhibition is achieved. It uses a repetition priming paradigm to examine whether effects are explained by a self-inhibitory process.
EXPERIMENTS 1 AND 2
In Expts 1 and 2, participants were asked to name a series of pictures, and we explored the relationship between the apparent inhibition of a particular response (indicated previously by analysis of naming errors) and naming latencies to a subsequent semantically related target trial. If a potential competitor to the target has been temporarily inhibited on the immediately preceding trial, then it follows that target naming latencies should be faster when a related prime and target are successive than when they are separated by a number of trials, when inhibition is no longer present. Consistent with this argument, Campbell (1990) found that the retrieval of answers to arithmetic problems was facilitated after supposed inhibition of a competitor.
One problem is that it can be difficult to examine both naming latencies and errors simultaneously. Speeded picture naming instructions do increase the error rate, but this can be at the expense of reliable or informative latencies. Instructions to name as quickly but as accurately as possible (standard naming instructions) clearly give better latency estimates, but it is not always the case that errors are sufficient for analysis. For these reasons, in Expt 1, speeded naming instructions were used, and errors were the main dependent variable. In Experiment 2, normal naming instructions were used, and naming latencies were the main dependent variable. The second dependent variable was noted in each case. Both experiments used the same stimuli and design, and so they are reported together.
In a design similar to one employed by Wheeldon and Monsell (1994), related primes and targets were presented, either in succession (Lag 1 condition), or separated by two intervening unrelated trials (referred to as Lag 3 condition, for consistency with our earlier work on lag analyses). A control condition was also included. All stimuli were pictures of objects. Previous work in our laboratory has indicated that inhibition has been removed sufficiently after a Lag of 3 for perseverative errors to occur as a result of the residual activation which remains in prime representations (Vitkovitch & Rutter, 2000). We predicted, therefore, that under speeded naming instructions (Expt 1), prime-related errors for targets under the Lag 1 condition should be significantly lower than in the Lag 3 condition, due to inhibition of the prime representations in the former condition. The control condition provided a baseline estimate of the occurrence of these same types of errors in the absence of a related prime. The Lag 3 error rate should be significantly higher than the control condition, while the Lag 1 error rate should be below that of the control condition. In Expt 2, in line with the semantic priming studies which examined target latencies (Wheeldon & Monsell, 1994), target pictures in the Lag 3 condition should have longer latencies than in the control conditions, as a result of specific interference from the prime. However, if prime representations are very briefly inhibited, then target picture naming latencies in the Lag 1 condition should be significantly less than in the Lag 3 condition. It is not clear whether latencies in the Lag 1 condition will also be longer than the control condition, because this could depend on whether all or only some aspects of the representation of that prime object are inhibited. Campbell (1990), for example, found latencies actually to be quicker in the related condition than in the unrelated condition.
Two groups of 24 students from the University of East London volunteered to take part in these experiments. All had English as their first language, and all reported normal or corrected eyesight. In both experiments, age ranged from 18 to 45 years, and the samples were approximately two-thirds female. In Expt 1, one participant who made no naming errors (apart from a Pass' response) was replaced.
Design and stimuli
A total of 36 pairs of related prime and target pictures were selected from the Snodgrass and Vanderwart (1980) set, from a range of semantic categories (e.g. animals, musical instruments, vehicles, clothes, household items, tools, fruits, vegetables). On the basis of inspection of naming errors from previous research, the prime was selected so that it appeared to be a strong competitor for the target. Examples of primes and targets are bus and lorry, guitar and violin, aeroplane and helicopter. The primes and targets were selected so that the different pairs did not interfere with each other. In addition to the prime and target, two unrelated filler pictures were selected, so that primes and targets were presented in sequences of four pictures. The lag between prime and target was manipulated, leading to two experimental conditions. In the Lag 3 condition, the two unrelated filler pictures intervened between prime and target (e.g. aeroplane, filler, fillet, helicopter). In the Lag 1 condition, the target imm ediately followed the prime, and the two unrelated fillers preceded the prime (filler, filler, aeroplane, helicopter). A third control condition was included, in which the target was presented without any related prime, but instead an extra unrelated picture was included (filler, filler, filler, helicopter). Thus, the sequences of four stimuli constituted one trial, although this was not apparent to participants. The 36 pairs of related prime and targets and fillers were divided into three lists of 12 pairs, such that there were an equal number of exemplars from each category in each list. The lists were rotated across the three conditions, so that, across the participant group, each target picture appeared under each of the three conditions an equal number of times. No stimulus ever appeared more than once for each participant.
The pictures were digitized and presented on a Macintosh computer using the Psychlab software package (Bub & Gum, 1990). Pictures were presented as black outline drawings within a light grey window. The picture disappeared from the screen after 500 ms and, in Expt 1, an auditory buzzer sounded at this point, to indicate a deadline which participants should try to beat. All participants were asked to name the pictures, but participants in Expt 1 were encouraged to try to do so before they heard the buzzer. They were advised nor to be concerned about any errors, since speed was more important than accuracy. Participants in Expt 2 were simply asked to respond as quickly but as accurately as they could. All participants wore a neckband with a microphone, which triggered a millisecond timer. After each response, there was a 4 second interval before the next picture was presented. The pictures were presented in a different random order to each participant, and the experimenter noted any hesitations, naming errors, and equipment failures. Participants were given four practice trials (12 stimuli) before the start of the experiment. They were informed that they could stop the experiment at any point if they so wished.
The results are organized so that the analyses of errors are presented first for both experiments, followed by analysis of naming latencies.
Analysis of errors
Naming errors were classified as prime-related when they corresponded to the related prime. in the control condition, a prime-related error was scored if the target error corresponded to the related (though unseen) prime for that specific target. The percentage of naming responses which were prime-related errors are presented for each condition in Table 1. There were also a number of errors which reflected the names of other exemplars from the same semantic category as the target, and for comparison purposes, the percentage error rates for these 'non-prime-related errors' are also presented in Table 1. Other inaccurate responses included the use of superordinate terms and the occasional 'Pass' response. There were very few unrelated responses (less than 1%).
The prime-related errors from Expt 1 were analysed using non-parametric statistics, because the data for the Lag 1 condition were not, nor were they expected to be, normally distributed. The Friedman analysis of variance by ranks indicated a significant difference across conditions, [[chi]sup.2](2) = 9.15, p 9.15, p = .01. Wilcoxon tests (one-tailed) were used for the follow-up comparison tests. As predicted, there were more prime-related errors in the Lag 3 condition than in the Lag 1 condition, Z = 3.09, p [less than] .01. For the two comparisons against the control condition, the alpha level was set at .025. There were significantly more prime-related errors in the Lag 3 condition than the control condition, Z = 2.43, p [less than] .01. The difference in error rate between Lag 1 and control condition approached significance, Z 1.60, p = .05.
The non-prime-related errors were analysed similarly using the Friedman test. Although there was some reduction in error rate in the Lag 3 condition, there were no significant differences between conditions, [[chi]sup.2](2) = 1.75, p [greater than] .05.
The prime-related error rare was low in Expt 2. Once again, though, a reduced error rate for the Lag 1 condition relative to the other two conditions is evident. Although the Friedman analysis showed no significant difference between the three conditions ([[chi].sup.2](2) = 1.71, p [greater than] .05), the planned Wilcoxon test indicated that the difference between Lag 3 and Lag 1 conditions was significant, Z = 1.77, p [less than] .05, (one-tailed test). The comparisons of each experimental condition against the control conditions did not reach significance (p [greater than] .025).
The non-prime-related errors failed to show any difference across conditions, ([[chi].sup.2](2) 0.44, p [greater than] .05.
Analysis of naming latencies
Median response latencies were calculated for each participant for each condition, excluding hesitations, or failures of the timing mechanism. Latencies were also excluded for any trials on which either the target or the prime had been misnamed. Table 1 gives the mean of the median response latencies for each condition for both Experiments. Unfortunately, latencies were not recorded on all trials for three participants in Expt 1, and so latencies were not available for these individuals. A further (outlier) participant was also eliminated due to very long latencies.
In all experiments, mean naming latencies were also calculated for each participant, but the results of these analyses (and any data transformations) will only be presented where these differ from analysis of median latencies.
A one-factor repeated-measures ANOVA indicated no significant differences in median latencies across conditions in Expt 1, P1(2, 38) = 1.03, p [greater than] .05, MSerror = 6427. However, in Expt 2, a significant difference between conditions was found, F1(2, 46) = 6.98, p [less than] .01, MSerror = 9704. The planned comparison of Lag 3 and Lag 1 latencies was significant, t(23) = 2.89,p [less than] .01, one-tailed, with longer target latencies in the Lag 3 condition. Comparisons of each experimental condition against the control condition, using Dunnett's modified t test, confirmed that latencies were longer in the Lag 3 condition than the control condition (p [less than] .01, one-tailed). There was no difference between the Lag 1 latencies and control latencies (p[greater than] .05). Analysis treating items as a random factor confirmed this pattern of results, P2(2, 68) = 7.67, p [less than] .01, Mserror = 13317. One item (tiger) was excluded from the analysis because there were insufficient latencies due to a high error rate. The differen ce between Lag 3 and Lag 1 condition was significant, t(34) = 3.83, p [less than] .01, one-tailed. Comparisons against the control condition (Dunnett's test) again indicated a difference between Lag 3 and the control condition (p [less than] .01, one-tailed test), but no difference between Lag 1 and control condition (p[greater than] .05).
The data from Expts 1 and 2 show a specific interference effect from semantically related prime pictures which have been presented three trials earlier. In Expt 1, the prime-related error rate in the Lag 3 condition was significantly higher than the control condition, and in Expt 2, Lag 3 target picture naming latencies were slower than the control condition. The analysis of latency data from Expt 1 did nor reveal any significant differences between the Lag 3 and control conditions. As noted earlier, this was not unexpected. In both experiments, however, the pattern of data for the secondary dependent variables was consistent with the results just reported, although in Expt 2 there was only a very slight rise in the Lag 3 prime-related error rate relative to the control condition. The most straightforward interpretation of the results for these two conditions across the experiments is that in the speeded naming condition, participants respond before they can eliminate the strong competition from the prime, an d so prime-related errors emerge in the Lag 3 condition. In Expt 2, where it is clear that they take more time to respond and make fewer errors generally, participants rarely make prime-related errors in the Lag 3 condition because they spend additional time in overcoming the competition from the prime. The semantic interference effect evident in the Lag 3 condition replicates other results reported in the literature which indicate that name retrieval involves competition from other exemplars from the same semantic category (e.g. Wheeldon & Monsell, 1994).
In both experiments, the prime-related error rate was significantly lower in the Lag 1 condition than the Lag 3 condition. Lag 1 prime-related error rate was also lower than the baseline estimate provided by the control condition, although this comparison of prime--related errors only approached significance in Expt 1; it may be difficult to establish a significant difference between these two conditions because of a floor effect. The differential prime-related error rate as a function of lag between prime and target is entirely consistent with our previous work on errors (Vitkovitch etal., 1996; Vitkovitch & Rutrer, 2000). These error data suggest that there may be a temporary inhibitory influence directed at prime representations which immediately precede a target, which limits the interfering potential of the prime. The latency data in Expt 2 are consistent with this interpretation. Earlier, we argued that if the potential for the prime to act as a competitor has been reduced in the Lag 1 condition, then naming latencies in the Lag 1 condition should be faster than latencies in the Lag 3 condition. This result was confirmed in Expt 2. Furthermore, naming latencies in the Lag 1 condition did not differ from the control condition.
There is, however, an alternative explanation for the reduced interference effects in the Lag 1 conditions. Wheeldon and Monsell (1994) similarly found that when related prime definitions and target pictures were separated by two intervening unrelated trials, the interfering effect was stronger than when prime definition and target picture followed each other. They suggested that at a Lag of 1, two counteracting influences were in operation. The interference effect from the prime was reduced, not because of suppression, but because of the prime resulting in a brief facilitatory effect. Facilitatory priming effects between related prime and target pictures have frequently been found in the literature (Bajo, 1988; Carr, McCauley, Sperber, & Parmalee, 1982), and the facilitation is usually explained by reference to automatic spreading activation within semantics, allowing pre-activation of target representations. Usually, though, the interval between prime and target is relatively short (e.g. less than 1 second ). However, semantic facilitation could occur at longer time intervals as a result of more deliberate, controlled processing (Neely, 1977). Thus, the explanations differ for what appears to be a similar result which has been found in two paradigms which differ mainly in the presentation form of the prime (definition or picture).
Given the similarity of the results across these studies, we should consider whether a facilitatory effect from the related prime in Lag 1 position can explain the current results. Anderson and Spellman (1995) have highlighted the difficulty in drawing inferences about inhibitory effects from response latencies in particular. However, an analysis of errors can help in the present case. The relative incidence of non-primerelated errors across conditions provides an indication of the competition from other category exemplars during name retrieval. To address the facilitatory account of reduced interference effects, we need to compare the incidence of non-prime-related errors across Lag 1 and control conditions within each experiment. If the reduced Lag 1 interference effects evident in the prime-related error analysis of Expt 1 and the latency analysis of Expt 2 were due to facilitation from the prime, then the non-prime-related errors should also be reduced in the Lag 1 condition. In Expt 1, the non-prime-rel ated error rate is at or just above 5% for both Lag 1 and control conditions, and there was clearly no significant difference between the two conditions. In Expt 2, although admittedly the error rate is really too low for meaningful interpretation (and there were no statistical differences in the overall analyses), it is again the case that there is a greater decrease in prime-related errors across control and Lag 1 conditions than there is in non-prime-related errors. Given that in Expt 1, the reduction in prime-related errors in the Lag 1 condition relative to the control approached significance, we maintain that the present data fit better with the suggestion that the absence of an interference effect from primes in Lag 1 condition is due to a brief suppression effect directed specifically at prime representations. We return to discussion of these two interpretations of reduced interference effects later.
There could be at least two mechanisms by which prime representations are inhibited. One possibility is that the Lag 1 prime, as a potentially very strong competitor, is subject to lateral inhibition from the related target itself. Dell and O'Segahdha (1994) discuss the possibility that lateral inhibition may be sufficiently dynamic to be directed to the most potent competitor, which could account for inhibition of an immediately preceding prime trial, but not one several trials earlier. A second possibility is that prime representations undergo a brief period of self-inhibition. Arbuthnott (1995) favours a self-inhibitory account of the inhibition which occurs in number-fact retrieval, and this also fits well with Houghton and Tipper's (1996) discussion of the use of inhibition in sequential tasks.
In the sequential naming task, if the object representation inhibits itself, rather than receives lateral inhibitory input from another competing representation, then it follows that there may be some difficulty in re-activating that same representation when it is re-presented immediately, in place of a semantically related trial. Experiment 3 investigates this possibility by examining whether there is any difference in target picture naming latencies as a result of retrieving the same name (during a prime trial) either immediately before the target, or three trials earlier.
There is ample evidence that this kind of repetition priming actually leads to strong facilitation in retrieving target names (Brown, Neblert, Jones & Mitchell, 1991; Dean & Young, 1996; Durso & Johnson, 1979; Ferrand, Grainger & Segui, 1994; Griffin & Bock, 1998 (Expt 1); Mitchell & Brown, 1988; Warren & Morton, 1982; Wheeldon & Monsell, 1992). One account of repetition priming suggests that there is a change in activation levels of some (or all) of the representations which are involved in picture naming, so that there is a benefit in repeated processing e.g. excitation may remain in structural representations (Dean & Young, 1996; Warren & Morton, 1982), or there may be a strengthening of the links between lexical representations (Monsell, Matthews, & Miller, 1992). Facilitation effects can last over a number of lags, and even over a number of weeks (Mitchell & Brown, 1988). Therefore, relative to baseline trials, we would expect some facilitation to occur in trials which repeat stimuli after two interveni ng stimuli (Lag 3 condition). There is also some evidence in other studies that facilitation is found when immediate repetition conditions are compared to control conditions (Arburhnott & Campbell, 1996; Durso & Johnson, 1979). This does not rule out the possibility of selfinhibition; suppression may be released when the same stimulus is re-presented (see, for example, Klein & Taylor, 1994; Neill, Valdes, & Terry, 1995; Tipper, Weaver, Cameron, Brehaut, & Bastedo, 1991) and/or self-inhibitory influences may be directed at only part of the prime representation, leaving other aspects highly activated. What would be of particular interest, however, would be slower target naming larencies in the immediate repetition condition relative to the Lag 3 repetition condition (our earlier experiments have shown that by Lag 3, representations have already recovered from the effects of inhibition). Such a pattern of results would be consistent with prime representations undergoing a brief period of self-inhibition. By cont rast, equivalent Lag 3 and Lag 1 repetition priming effects (or slightly stronger Lag 1 effects) might be expected if there were no temporary period of self-suppression.
Experiment 3 uses the same design as the two previous experiments, although some changes were introduced in an attempt to overcome certain difficulties inherent in repetition priming studies. The prime stimuli were presented as definitions, from which the object name had to be retrieved. Wheeldon and Monsell (1992, 1994) used a method of alternating definitions and pictures to minimize the possibility of retrieval of an episodic trace, which can occur when participants detect similarity between prime and target trial (see Brown et al., 1991; Dean & Young, 1996; Jacoby, 1983; Monsell, 1991; Neill, 1997; Wheeldon & Monsell, 1992, for a more detailed discussion of episodic and other, related accounts of repetition priming). The use of an episodic trace might be particularly likely when trials are repeated in succession, and so in the present context, might allow participants to simply restate the previous name, and possibly by-pass any self-inhibited representations. Wheeldon and Monsell (1992) suggest that cha nging the actual form and task requirements for prime and target would reduce the likelihood that participants would retrieve an episodic trace, and encourage object name retrieval via access to stored representations. They argued that the processes involved in retrieving the name from a definition and naming a picture should overlap at the stage of activating semantic representations, a stage which is prior to lexical retrieval.
A second potential problem is that inhibition during sequential picture naming may be under strategic control, and that participants either do or do not inhibit a response, depending on whether they perceive it to be useful for later trials. So, in blocks of trials where stimuli are frequently repeated, self-inhibitory mechanisms may simply be abandoned. Arbuthnott and Campbell (1996) have examined whether negative error priming in arithmetic retrieval is due to intentional suppression, but did find some evidence for automatic inhibitory processes. However, in the present case of object naming, we cannot rule out the possibility that immediate repetition of trials may lead participants to abandon the use of inhibitory mechanisms which under other circumstances they make use of.
Arguments for or against self-inhibitory mechanisms would therefore be more compelling if repetition priming data could be evaluated in the context of other data which were consistent with inhibitory effects. Therefore, in Expt 3, we aimed to replicate the reduced semantic interference effects from Lag 1 primes which were evident in the naming latency data of Expt 2. We included semantic priming trials in addition to repetition priming trials.
Experiments which include both repetition and semantic priming trials are not without their own difficulties, though, and Neumann (2001) has highlighted research which suggests that strategies induced by one condition can be applied inappropriately to other conditions, distorting results or interpretation. In studies of selective attention, negative priming is apparent when an unattended prime is presented as a target on a subsequent trial. Positive (facilitatory) priming can also be found when the attended prime is repeated on the next trial. Neumann suggests that the inclusion of repeated items in negative priming studies might allow participants to develop anticipatory strategies, which would have adverse effects for trials where stimuli are not repeated. The size of repetition priming effects may depend on the proportion of repeated trials, and discrepancies across studies may be due to this. Conceivably, also, the size of negative priming effects may be influenced by the inclusion of repetition priming trials. Arbuthnott and Campbell (1996), for example, have suggested that the inclusion of repeated trials in studies of negative error priming in number fact retrieval may lead participants to develop 'response detection' strategies, which might mask negative error priming effects. As in the episodic retrieval account of repetition priming, they suggest that participants' awareness of repetition may lead them to check the last response to see if it matched the requited target name, and if so, they respond with this answer. Arbuthnott and Campbell (1996) showed that participants use of the prime is dependent on their detecting similarity between repeated trials, and mistakenly detecting similarity in related trials (e.g. 4 X 6, 4 X 8). Using this strategy, participants would, for some trials, by-pass the usual retrieval of arithmetic facts from stored knowledge. A strategy such as this would be likely to lead to occasional errors and longer latencies on trials which were immediately preceded by primes which cl osely resembled the target in form, and this is what they found.
Therefore, there does need to be caution in interpreting results from repetition priming studies, and also those which have both related and repetition prime conditions. For this reason, Expt 3 represents only a preliminary attempt to investigate the issue of self-inhibitory mechanisms. Nevertheless, we considered it worth running a repetition priming experiment, since the finding of stronger Lag 3 facilitatory priming effects (relative to Lag 1 effects) would be good evidence for a self-inhibitory mechanism. There is a difficulty, however, in settling on an appropriate number of repetition trials; the work by Neumann (2001) has shown that participants can be biased towards or against repetition, influencing the results. In the following experiment, we considered it important to demonstrate inhibitory effects in the semantic priming trials, and to avoid strategic use of the prime, and so we aimed to avoid a bias towards repetition. The proportion of repetition trials relative to non-repetition trials was kep t low. In the Arbuthnott and Campbell (1996) study, for example, 25% of trials were immediate repetitions, and negative error priming was evident at least when primes were dissimilar to targets. In the following experiment, repetition primes were also presented on approximately 25% of trials (where trials refers to sequences of four stimuli). Although the actual number of repetition priming trials was only six for each condition, repetition priming effects in object naming are generally robust, and we anticipated that they would be detected with this number of trials.
To review, relative to the control condition, we expected to replicate the previous finding of interference from semantically related definition primes in the Lag 3 condition (Wheeldon & Monsell, 1994), and to find reduced or no interference effect from related primes in the Lag 1 condition, suggesting a brief inhibitory effect. If this were accompanied by reduced repetition priming effects for the Lag 1 condition relative to the Lag 3 repetition priming condition, then this pattern of latency data as a whole would be consistent with an inhibitory effect occurring as a result of a self-inhibitory mechanism, whether automatic or intentional. If, on the other hand, reduced Lag 1 interference effects were found in tandem with equivalent Lag 1 and Lag 3 repetition priming, then this would suggest that inhibition in the Lag 1 semantic trials was accomplished by some mechanism other than self-inhibition. By contrast, if the inclusion of repeated trials results in participants abandoning the use of inhibition, or i nduces strategic use of the prime (e.g. anticipatory strategies or a bias towards repetition, or retrieval of episodic trace), then we might expect greater facilitatory priming in the Lag 1 repetition priming condition relative to the Lag 3 condition coupled with costs for the semantic priming conditions; in particular, the Lag 1 related condition would then show stronger interference effects from Lag 1 primes than from Lag 3 primes.
A total of 36 volunteers from the same population of students as those participating in the previous experiments were tested. They met the same criteria as before.
Design and stimuli
The basic design was as for Expts 1 and 2, in that primes were either presented in Lag 3 or Lag 1 position, and control conditions were again incorporated. Unrelated filler stimuli were included as before. However, in this experiment both repetition priming and semantic priming were examined simultaneously. All primes were presented as definitions, from which the object name had to be retrieved. The target was always a picture. On some trials (approximately 25% - see below), the name retrieved from the prime definition was the same as the name to be retrieved from the target picture. However, for approximately 40% of the trials, the prime was semantically related to the target, as in Expts 1 and 2. The remaining trials were unprimed.
For the Lag 3 condition, the sequence of stimuli was; prime definition, filler picture, filler definition, target picture (the target could be either the same or semantically related to the prime). In the Lag 1 condition, the sequence was: filler definition, filler picture, prime definition, target picture (again, repetition or related). The two control conditions (one to allow comparisons with the related prime conditions, and one for comparisons with repetition priming conditions) consisted of three unrelated filler stimuli before the target, with the same sequencing of definitions and pictures.
Some changes were made to the actual lists of prime and target stimuli from the previous experiments, although selection was from similar categories. This was largely dictated by the success of definitions in eliciting the intended name. Definitions generally included the superordinate term and a description of the object (e.g. skirt - item of clothing for women which hangs from the waist). Where possible, the object was defined by the use of functional features, but on some occasions, definitions included visual features. A list of potential prime and filler definitions was pre-tested on a group of 10 participants, and a criterion of 80% correct was adopted. Definitions which were unsuccessful were altered or replaced. As before, target and filler pictures were selected from the Snodgrass and Vanderwart (1980) set.
In all, 30 semantically related prime and target pairs were prepared, and these were divided into three lists of 10 pairs, which were rotated across the three semantic priming conditions as before (Lag 3, Lag 1, unprimed control). For the repetition priming conditions, a separate set of 18 prime and target pairs were selected (from similar semantic categories), such that the prime successfully elicited the target picture name. These were divided into three lists of six prime and target pairs, which were also rotated across the repetition priming conditions (Lag 3, Lag 1, unprimed control). A full list of prime definitions and target pictures is given in the Appendix.
Definitions and pictures were presented as before in the centre of a window on the screen of a Macintosh computer. Definitions occupied between two and three lines, and the whole definition was presented at once. The participants were asked to retrieve the object name, either from a definition, or a picture, as quickly but as accurately as they could. The stimulus remained on the screen until the response, and then the screen remained blank for 3 seconds. Following this, a fixation symbol appeared for one second, which reminded participants as to the nature of the next stimulus; a picture was preceded by the symbol [ ], and a definition was preceded by a series of dashes: - - -. Participants received four practice trials (12 stimuli) before the experimental trials.
The sequences of four stimuli were randomized separately for each participant. Naming latencies, measured to the nearest ins, were automatically recorded, while the experimenter noted any errors, hesitations or equipment failures.
The data were variable, with a few latencies even longer than 3 seconds. Reciprocal transformations were used to minimize the influence of outlier participants. In the calculation of participant means, target latencies above 3 seconds were excluded, providing they were more than 2 standard deviations (SD) above a participant's mean. This led to the exclusion of only six scores. As before, results of analysis on means and transformations will only be reported where these differ from the results of median analysis.
Median participant response latencies were calculated for all six conditions, after exclusion of errors and machine failures as before. The means of medians (and percentage error rates) are shown in Table 2. One participant's median latency for the Lag 1 semantic priming condition was identified as an extreme point, using exploratory data analysis techniques, substantially influencing the mean value, and Table 2 shows the means for the semantic priming conditions excluding this participant's data. Note though that analyses are conducted both with and without this participant's response.
It is clear that the pattern of data for repetition priming and semantic priming conditions are different. Although it was nor the original intention to analyse these data together, an initial analysis which included type of priming as a factor (semantic or repetition) in addition to prime condition (Lag 3, Lag 1 or unprimed) indicated a significant interaction between the two factors, F(2, 70) = 3.28, p [less than] .05 MSerror = 30138).
The pattern of data for the semantic priming conditions does show that the interference from related primes is less marked for the Lag 1 condition than for the Lag 3 condition. However, the planned comparison between these two conditions only approached significance in the items analysis (by participants, t(35) = 0.10, p [greater than] .10; by items, t(27) = 1.58, p = .06, one-tailed). Two items (hen and peach) were excluded from the analysis because of high error rates. However, comparisons against the control condition, using Dunnett's test, indicated that the Lag 3 latencies were significantly longer than latencies in the control condition, for both participant and item analyses, p [less than] .05, one-tailed test). Latencies in the Lag 1 condition did not differ from the control condition (p [less than] .05). These same results were obtained when reciprocal transformations were used, or when the participant with extreme Lag 1 latencies was removed from the analysis.
Using the Friedman analysis, no difference was found in the prime-related error rate across conditions, [[chi].sup.2] = 0.54, p [less than] .05. Similarly, there was no difference in nonprime-related errors, [[chi].sup.2](2) = .09, p [less than] .05.
Although the means in Table 2 show that target pictures were named more quickly in the two prime conditions than in the control condition, comparisons of median latencies, using Dunnett's test, indicated no significant differences, either by participants or by items (p [less than] .05, one-tailed). The planned comparison of Lag 1 latencies against Lag 3 latencies was also not significant, by participants, t(35) = 1.34, p [less than] .05, and by items, t(17) = 0.83, p [less than] .05.
Analysis of participants' mean latencies, however, did show that latencies were faster in the Lag 1 condition than in the control condition (using Dunnett's test, p [less than] .05, for both participants and items analyses (one-tailed)). Using a reciprocal transformation, comparisons of Lag 3 mean latencies against the control approached significance (p [less than] .10 (one-tailed)) for both items and participants.
Analysis of errors (total errors) indicated no significant difference between Lag 1 and Lag 3 conditions, t(35) = 0.85, p [less than] .05. Comparisons against the control conditions, using Dunnett's test, resulted in a significant difference between the Lag 1 and control condition (p [less than] .01 (one-tailed)). The difference between Lag 3 and control condition just failed to reach significance at the 5% level (p [less than] .10, one-tailed).
The data from Expt 3 were noisy, and effects were not as clear-cut as in the previous two experiments. There may be three reasons for this. First, participants may have needed more practice at retrieving names from definitions, and switching between picture and definition trials. Second, the number of trials per cell was also reduced for the repetition priming conditions. Finally, the inclusion of both repetition and interference trials may have tempered effects (Neumann, 2001), although this is difficult to assess from this single experiment. Nevertheless, the pattern of data across conditions allows us to draw some preliminary conclusions concerning the issue of self-inhibitory mechanisms during picture name retrieval.
We argued that if there is a brief inhibitory mechanism operating during picture naming, then Lag 3 related primes should interfere with retrieving the names of target pictures, but Lag 1 primes should not. This pattern of data was evident; target latencies in the semantic priming Lag 3 condition were significantly longer than in the control condition, while there was no difference between Lag 1 and control target latencies. We discuss other aspects of the semantic priming results shortly, but the pattern of data is atleast consistent with the suggestion that primes in Lag 1 condition are temporarily inhibited, so that they do not interfere with retrieving target names. Although we cannot rule out the possibility that the inclusion of repetition trials may have had some influence on processing in other conditions, certainly the data do not suggest that inhibitory mechanisms have simply been abandoned, or that participants made consistent strategic use of the prime throughout the experiment (either in an anti cipatory or an episodic fashion). Were this the case, the semantic priming data would have shown specific costs in the Lag 1 condition.
Turning now to the repetition priming data, we predicted that if the inferred inhibition operates on a self-inhibitory basis, then facilitatory priming effects should be stronger for Lag 3 primes than for Lag 1 primes. This pattern of results was not evident. Target latencies in the Lag 1 and Lag 3 conditions did not differ significantly, and naming latencies were in fact slightly faster in the Lag 1 condition. Latencies in the Lag 1 repetition priming condition were significantly faster than the control condition when the analysis was conducted on means, suggesting that effects may have been restricted to slower responses. The comparison of Lag 3 and control condition mean target latencies only approached significance using Dunnett's test. Analyses of errors were consistent with the latency data; there were significantly less errors in the Lag 1 condition than in the control condition, and the difference between Lag 3 and control conditions just failed to reach significance using Dunnett's test. Therefore, the repetition priming data show some indication of facilitatory priming, but this was only really evident in the Lag 1 condition. This is not consistent with the suggestion that primes in the semantic and repetition priming Lag 1 condition undergo a brief suppression as a result of an automatic self-inhibitory mechanism. If the lack of interference of Lag 1 primes in the semantic conditions does reflect temporary inhibition, then the data suggest that it may be accomplished by a method other than self-inhibition. Inhibition appears to be restricted to trials where a prime is followed by a competing target (semantically related), suggesting that inhibition occurs as a result of processing the competitor target. A possible candidate mechanism is that of lateral inhibition from the activated target representations to the prime representations. However, if this occurs, any such inhibition must be both fast acting, and must be more strongly directed to the representations of a very recently named competitor (i.e. the semantically related Lag 1 prime), than to a competitor named three trials earlier. Following through the argument for lateral inhibition; in the case of the repetition priming trials, there is no related target to exert an inhibitory influence on the Lag 1 prime representations, and so some facilitation is evident as a result of the repeated processing.
One worrying aspect of the repetition data is that facilitation priming effects in the Lag 3 condition did not reach statistical significance. Durso and Johnson (1979) did not find that facilitatory effects reduced markedly over 2 to 8 lags, and Wheeldon and Monsell (1992) found repetition priming from retrieving object names from definitions several trials earlier. We need therefore to consider whether the higher proportion of non-repetition trials actually biased participants' expectancies against repetition, overriding to some extent any automatic component to the facilitatory priming (Neumann, 2001). An experiment altering the proportion of semantic and repetition priming trials could address this possibility directly. However, any argument for a bias against repetition in the present experiment needs to accommodate the finding that facilitatory priming was mainly evident in the Lag 1 condition. Either it is the case that any bias against repetition is stronger for Lag 3 repetition (which seems unlikely, and does not fit the Lag 3 semantic interference effects), or else we need to acknowledge that, despite any such bias, the data are still inconsistent with the suggestion that representations undergo a brief period of self-inhibition.
We earlier argued that the semantic priming data are not consistent with routine episodic evaluation of the prime as a candidate target response on every single trial. However, it does remain a possibility that the Lag 1 repetition priming effect is due to strategic episodic use of the prime which is restricted mainly to the repetition priming trials. This would imply that participants only use the prime after a relatively late stage of processing the target, when they are reasonably confident that prime and target match. For example, participants could fully process the target semantically, and only then, if this matched the prime semantic specification, by-pass aspects of target name retrieval by simply re-stating the prime name from the immediately preceding trial. This might lead to the occasional error or increased latency when prime and target resembled each other very closely in the semantic priming conditions. There is some indication of this in the slightly raised prime-related error rate for Lag 1 semantic priming condition; the error rate is actually higher than the control, and this is due largely to one set of highly similar prime and target objects (hen misnamed as prime cockerel).
In summary, the semantic priming data from Expt 3 show a pattern which is generally consistent with the existence of a brief inhibitory effect after name retrieval. The pattern of data from repetition trials is not consistent with any such inhibitory effect operating as a result of a self-inhibitory mechanism. There was no indication of any specific difficulty in immediately retrieving the same name, and in fact facilitatory priming was mainly indicated only in the immediate repetition trials. The patterns of semantic and repetition priming data, taken together, do not indicate any clear-cut bias either towards or against repetition, though we cannot rule out the possibility that the inclusion of companion conditions may have to some extent influenced interference or facilitatory effects. Similarly, the overall pattern of data suggests that there is no systematic episodic use of the prime. But again, we cannot eliminate the possibility that strategic episodic use of the prime response in repetition trials on ly may have allowed participants to by-pass selfinhibited lexical representations.
Earlier research in our laboratory has shown that when participants name semantically related pictures under speeded naming instructions, they make errors which relate to earlier named pictures but not the immediately preceding picture (Vitkovitch et at., 1996; Vitkovitch & Rutter, 2000). We have interpreted this as evidence that sequential picture naming involves a brief suppression of the just-named picture representations. Two of the three experiments reported here have examined the relationship between the apparent inhibition of one naming trial, and naming latencies on the next trial. Experiments 1 and 2 used exactly the same design and stimuli, and complemented each other by providing error data and latency data which showed that there was no interference from related primes when they were presented immediately before a target picture, although interference was evident from a prime presented three trials earlier. The error data from Expt 1 are therefore consistent with our own earlier work, and are usef ul because we are now able to show negative error priming across two different designs -- one in which naturally occurring errors are analysed as a function of lag between error and earlier trial, and the other where the lag between prime and target stimuli is actually manipulated.
Yet further support for reduced interference effects from Lag 1 primes comes from the paradigm used in Expt 3. Here, primes in Lag 1 or Lag 3 conditions were presented as definitions, from which the object name had to be retrieved. Target naming latencies were longer in the Lag 3 related condition than the control condition. There was no significant difference between Lag 1 and control conditions.
In summary, Expts 1, 2 and 3 all show reduced interference effects from Lag 1 related primes relative to Lag 3 related primes. We have interpreted this as evidence for a brief inhibitory effect, but we referred to another possible explanation of the data. Wheeldon and Monsell (1994), in their definition priming experiment, suggested that a facilitatory priming effect in the Lag 1 condition counteracted the competitor priming effect, though the evidence they present in favour of this is indirect. We argued against this in Expts 1 and 2, because the data did show a specific reduction of prime-related errors in particular -- other errors which reflected the names of exemplars from the same semantic category were not significantly reduced in the Lag 1 condition. Furthermore, in our work on the analysis of naturally occurring errors as a function of lag between error and earlier trial (Vickovitch et al., 1996; Vitkovitch & Rutter, 2000), we do not have this ambiguity in interpretation. In this paradigm, we are ab le to show that a particular trial might be subject to interference from a response three or four trials earlier, but at the same time there is specifically no interference from the immediately preceding trial. This pattern of data cannot be accounted for by facilitatory priming from the immediately preceding trial. Therefore, we are confident that an inhibitory account fits the picture naming data in Expts 1 and 2 more comfortably than an account which includes the operation of dual processes of competitor and facilitatory priming. In Expt 3, which was most similar to Wheeldon and Monsell's (1994) study, our arguments for ruling our faciliratory priming are perhaps less convincing, though again the error data are not consistent with such an account. Retrieving an object name from a definition rather than a picture is quite likely to encourage a richer semantic processing than retrieval of a name from a picture, and it remains a possibility that facilirarory priming might be occurring in Expt 3 but not in Exp ts 1 and 2. On the grounds of parsimony, however, we prefer an explanation which encompasses the similar results from the all three experiments and an explanation which is based on the operation of a single process rather than two opposing processes. On this basis, then, we would argue that there is little or no interference from primes which are presented immediately before a semantically related target because the prime representations undergo a brief period of inhibition. Such a conclusion fits well with other research on inhibitory effects in cognition, and in particular both the data and interpretations parallel findings by Campbell and colleagues in their study of number fact retrieval (Campbell, 1990; Campbell & Clark, 1989).
Experiment 3 also included repetition priming trials to allow a preliminary investigation of whether any such inhibition in sequential picture naming is the result of a self-inhibitory mechanism. We predicted that facilitatory priming should be stronger in the Lag 3 repetition condition than the Lag 1 condition if prime representations suppressed themselves very briefly. We found no evidence in this experiment that there was a specific difficulty in immediately re-activating prime representations; evidence for facilitatory priming was most apparent when the repetition trials were in immediate succession (Lag 1 condition). This suggests that the inhibition we have argued for in the semantic priming Lag 1 trials may occur as a result of another mechanism, such as lateral inhibition from the target representations to the competing related prime representations. The results from the present experiments cannot address the issue of alternative mechanisms of inhibition directly, although we did note that any latera l inhibition from the target to the prime representations must be both immediate and most potently directed to the strongest competitor.
There are several provisos to the conclusion that a self-inhibitory mechanism does not account for the reduced Lag 1 semantic interference effects. First, repetition priming effects were not strong, and were barely evident in the Lag 3 conditions. We noted a number of possible reasons for this, including the possibility that the inclusion of semantic priming trials may have tempered repetition effects. Second, an episodic retrieval account would fit data which showed stronger immediate repetition priming effects than for earlier trials. We are confident that participants are not routinely evaluating the Lag 1 prime as a candidate response to the target, because such a strategy would lead to specific costs in the Lag 1 semantic priming conditions. However, participants might restrict themselves to using the prime response only after they are confident that the target and prime match, and if so, they may by-pass lexical/ phonological representations which have been self-inhibited. Some accounts of language pro duction do include self-inhibitory mechanisms which are directed at phonological representations (e.g. Dell & O'Seaghdha, 1991; Mackay, 1987). In other work, we are addressing the locus of inhibition during picture naming, and it would clearly be useful to combine this with the current work on the mechanism of inhibition.
In conclusion, the experiments reported here have provided data consistent with earlier work on inhibitory effects during sequential object naming, and have allowed a preliminary investigation into the mechanism of this inhibition. The data from the repetition priming conditions in Expt 3 show that participants do not have difficulty in immediately re-activating a just-named object, despite data in the same experiment which suggests that under other conditions, just-named objects are briefly inhibited. Although the data are not consistent with suppression occurring as a result of self-inhibitory mechanisms, we acknowledge that there is more than one interpretation of the repetition priming data, and that further work is required before such a mechanism is ruled out.
This work was supported by an ESRC grant awarded to the first author (R000221593).
(*.) Requests for reprints should be addressed to Dr Melanie Vitkovitch, Department of Psychology, University of East London, Romford Road, London E15 412, UK (e-mail: email@example.com).
Anderson, M. C., & Bjork, R. A. (1994). Mechanisms of inhibition in Long-term memory: A new taxonomy. In D. Dagenbach & T. H. Cart (Eds.), Inhibitory processes in attention, memory, and language (pp. 265--325). New York: Academic Press.
Anderson, M. C., & Spellman, B. A. (1995). On the status of inhibitory mechanisms in cognition: Memory retrieval as a model case. Psychological Review, 102, 68-100.
Arbuthnott, K. D. (1995). Inhibitory mechanisms in cognition: Phenomena and models. Current Psychology of Cognition, 14, 3-45.
Arbuthnott, K. D. (1996). To repeat or not to repeat: Repetition facilitation and inhibition in sequential retrieval. Journal of Experimental Psychology: General, 125, 261-283.
Arbuthnott, K. D., & Campbell, J. I. D. (1996). Effects of operand order and problem repetition or error priming in cognitive arithmetic. Canadian Journal of Experimental Psychology, 50, 182-196.
Bajo, M. T. (1988). Semantic facilitation with pictures and words. Journal of Experimental Psychology: Learning, Memory and Cognition, 14, 579-589.
Brown, A. S. (1981). Inhibition in cued retrieval. Journal of Experimental Psychology: Human Learning and Memory, 7, 204-215.
Brown, A. S., Neblett, D. R., Jones, T. C., & Mitchell, D. B. (1991). Transfer of processing in repetition priming: Some inappropriate findings. Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 514-525.
Bub, D., & Gum, T. (1990). Psychlab (version 0.85). (Computer programme available from D. Bub, 3815 Oxford Avenue, Montreal, PQH4A 2Y3).
Campbell, J. I. D. (1990). Retrieval inhibition and interference in cognitive arithmetic. Canadian Journal of Psychology, 44, 445-464.
Campbell, J. I. D. (1991). Conditions of error priming in number-fact retrieval. Memory and Cognition, 19, 197-209.
Campbell, J. I. D., & Arbuthnott, K. D. (1996). Inhibitory processes in sequential retrieval: Evidence from variable-lag repetition priming. Brain and Cognition, 30, 59-80.
Campbell, J. I. D., & Clark, J. M. (1989). Time course of error priming in number-fact retrieval: Evidence for excitatory and inhibitory mechanisms. Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 920-929.
Carr, T. H., McCauley, C., Sperber, R. D., & Parmalee, C. M. (1982). Words, pictures, and priming: On semantic activation, conscious identification, and the automaticity of information processing. Journal of Experimental Psychology: Human Perception and Performance, 8, 757-777.
Dean, M. P., & Young, A. W. (1996). An item-specific locus of repetition priming. The Quarterly Journal of Experimental Psychology, 49A, 269-294.
Dell, G. S., & O'Seaghdha, P. G. (1991). Mediated and convergent lexical priming in language production: A comment on Levelt et al. (1991). Psychological Review, 98, 604-614.
Dell, G. S., & O'Seaghdha, P. G. (1994). Inhibition in interactive activation models of linguistic selection and sequencing. In D. Dagenbach & T. H. Carr (Eds.), Inhibitory processes in attention, memory, and language (pp. 409-453). New York: Academic Press.
Durso, F. T., & Johnson, M. K. (1979). Facilitation in naming and categorizing repeated pictures and words. Journal of Experimental Psychology: Learning, Memory and Cognition, 5, 449-459.
Ferrand, L., Grainger, J., & Segui, J. (1994). A study of masked form priming in picture and word naming. Memory and Cognition, 22, 431-441.
Gernsbacher, M. A,, & Faust, M. E. (1995). Skilled suppression. In F. N. Dempster & C. J. Brainerd (Eds.), interference and inhibition in cognition (pp. 295-3 27). San Diego: Academic Press.
Glaser, W. R., & Glaser, M. O. (1989). Context effects in Stroop-like word and picture processing. Journal of Experimental Psychology: Human Perception and Performance, 118, 13-42.
Griffin, Z. M., & Bock, K. (1998). Constraint, word frequency and the relationship between lexical processing levels in spoken word production. Journal of Memory and Language, 38, 313-338.
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). New York: Academic Press.
Houghton, G., & Tipper, S. P. (1996). Inhibitory mechanisms of neural and cognitive control: Applications to selective attention and sequential action. Brain and Cognition, 30, 20-43.
Humphreys, G. W., Riddoch, M. 3., & Quinlan, P. T. (1988). Cascade processes in picture identification. Cognitive Neuropsychology, 5, 67-103.
Humphreys, G. W., Lloyd-Jones, T. J., & Fias, W. (1995). Semantic interference effects on naming using a postcue procedure: Tapping the links between semantics and phonology with pictures and words. Journal of Experimental Psychology: Learning, Memory and Cognition, 21, 961-980.
Jacoby, L. L. (1983). Perceptual enhancement: Persistent effects of an experience. Journal of Experimental Psychology: Learning, Memory and Cognition, 9, 21-38.
Klein, R. M., & Taylor, T. L. (1994). Categories of cognitive inhibition with reference to attention. In D. Dagenbach & T. H. Cart (Eds.), Inhibitory processes in attention, memory, and language (pp. 113-150). New York: Academic Press.
Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual memory representations. Journal of Memory and Language, 33, 149-174.
McClelland, J. L., & Rumelhart, D. E. (1981). An interactive activation model of context effects in letter perception: 1. An account of basic findings. Psychological Review, 86, 287-330.
Mackay, D. G. (1987). Self-inhibition and the disruptive effects of internal and external feedback in skilled behaviour. In H. Heur & C. Fromm (Eds.), Generation and modulation of action patterns (Experimental Brain Research Series, No. 15, pp. 174-186). Betlin: Springer-Verlag.
Mitchell, D. B., & Brown, A. S. (1988). Persistent repetition priming in picture naming and its dissociation from recognition memory. Journal of Experimental Psychology: Learning, Memory and Cognition, 14, 213-222.
Monsell, S. (1991). The nature and locus of word frequency effects in reading. In D. Berner & G. W. Humphreys (Eds), Basic Processes in Reading: Visual word recognition. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Monsell, S., Matthews, G. H., & Miller, D. C. (1992). Repetition of lexicalisation across languages: A further test of locus of priming. Quarterly Journal of Experimental Psychology, 44A, 763-783.
Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Evidence for facilitatory and inhibitory processes. Journal of Experimental Psychology: General, 106, 226-254.
Neill, W. T. (1997). Episodic retrieval in negative priming and repetition. Journal of Experimental Psychology: Learning, Memory and Cognition, 23, 1291-1305.
Neill, W. T., 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: Academic Press.
Neumann, E. (2001). Cautions about asymmetric transfer in within-subjects experiments: Examples and evidence from recent selective attention investigations. Visual Cognition, 8.
Snodgrass, J. G., & Vanderwart, M. (1980). A standardised set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human, Learning and Memory, 6, 174-215.
Starreveldt, P. A., & La Heij, W (1995). Semantic interference, orthographic facilitation, and their interaction in naming tasks. Journal of Experimental Psychology: Learning Memory and Cognition, 21, 686-698.
Tipper, S. P., & Driver, J. (1988). Negative priming between pictures and words in a selective attention task: Evidence for semantic processing of ignored stimuli. Memory and Cognition, 16, 64-70.
Tipper, S. P., Weaver, B., Cameron, S., Brehaut, J. C., & Bastedo, J. (1991). Inhibitory mechanisms of attention in identification and localisarion tasks: Time course and disruption. Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 68 1-692.
Vitkovitch, M., & Humphreys, G. W. (1991). Perseverant naming errors in speeded picture naming: Its in the links. Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 664-680.
Vitkovitch, M., Kirby, A., & Tyrrell, L. (1996). Patterns of excitation and inhibition in picture naming. Visual Cognition, 3, 61-80.
Vitkovitch, M., & Rutter, C. (2000). The effects of response stimulus interval on error priming in sequential object naming. Visual Cognition, 7, 645-670.
Warren, C., & Morton, J. (1982). The effects of priming on picture recognition. British Journal of Psychology, 73, 117-129.
Wheeldon, L. R., & Monsell, S. (1992). The locus of repetition priming in spoken word production. The Quarterly Journal of Experimental Psychology, 44A, 723-761.
Wheeldon, L. R., & Monsell, S. (1994). Inhibition of spoken word production by priming a semantic competitor. Journal of Memory and Language, 33, 332-356.
Table 1. Mean naming latencies and percentage error rate for each semantic priming condition in Expt 1 (speeded naming instructions) and Expt 2 (standard naming instructions) Lag 3 Lag 1 Control Expt 1 Latencies Mean 581 547 552 (SD) (117) (106) (102) Error rate Prime-related 8.3% 1.4% 3.8% Non-prime-related 3.1% 5.2% 5.5% Expt 2 Latencies Mean 1056 968 960 (SD) (111) (131) (125) Error rate Prime-related 2.8% 0.7% 2.1% Non-prime-related 2.8% 1.0% 1.7% Table 2. Mean naming latencies and percentage error rate for semantic and repetition priming conditions in Expt 3 Lag 3 Lag 1 Control Semantic priming Latencies Mean 1203 1175 1138 (SD) (234) (193) (170) Error rate Prime-related 2.5% 3.6% 2.7% Non-prime-related 3.3% 3.3% 2.8% Repetition priming Latencies Mean 1163 1121 1193 (SD) (246) (195) (252) Error Rate 6.9% 5.1% 12.0% Appendix Prime and target stimulus lists for Expts 1 and 2 Prime Target List A horn trumpet bus lorry giraffe zebra teapot kettle hen duck boot shoe spoon fork cherry strawberry whale seal gun cannon scissors saw List B guitar violin aeroplane helicopter leopard tiger cup glass owl eagle skirt dress chair sofa onion pumpkin deer goat belt bow cake bread spanner pliers List C piano harp bicycle motorbike pig rhinocerous jug vase ostrich peacock shirt jumper fridge cooker apple pear dog fox comb brush ball balloon hammer axe Targets and prime definitions used in Expt 3 Semantic Priming Stimuli Target Prime List A Trumpet Horn Lorry Bus Zebra Giraffe Kettle Teapot Hen Cockerel Shoe Boot Strawberry Pineapple Screw Nail Cooker Fridge Goat Deer List B Violin Guitar Helicopter Aeroplane Leopard Tiger Glass Cup Swan Duck Dress Skirt Lemon Orange Screwdriver Pliers Fork Spoon Semantic Priming Stimuli Target Prime definition List A Trumpet Musical brass instrument, one type of which is called French. Lorry Vehicle with four wheels, run for the public, with a driver and conductor. Zebra Animal which is very tall and has a long neck and legs. Kettle Household container with a handle, spout and lid and in which a hot drink is brew Hen A bird which crows at the crack of dawn in the farmyard. Shoe Item of footwear which reaches above the ankle. Strawberry Fruit that is tropical, with a hard skin and leaves sprouting from the top. Screw Small sharp metal spike with a flattened head, which can be used to hang pictures Cooker Household appliance which keeps food cold. Goat Animal with hooves, the males of which have antlers. List B Violin Musical instrument that has six strings, and is played with fingers or a plectrum Helicopter Vehicle which flies and carries passengers. Leopard Animal which is a fierce member of the cat family, with stripes. Glass Household object which is used for drinking and has a handle. Swan Bird which is found on ponds, and which children feed bread to. Dress Item of clothing for women, which hangs from the waist. Lemon Fruit that is citrus, round, and can be made into squash or freshly squeezed. Screwdriver Tool that has pincers for holding small objects, and bending wire. Fork Household utensil used for stirring or for eating dessert. Fox Dog List C Harp Piano Motorbike Bicycle Rhino Elephant Vase Jug Peacock Ostrich Jumper Shirt Peach Cherry Spanner Chisel Rabbit Mouse Frying pan Rolling pin Repetition Priming Stimuli List A Tomato Tomato Sheep Sheep Toaster Toaster Belt Belt Beetle Beetle Saw Saw List B Potato Potato Donkey Donkey Saucepan Saucepan Fox Animal that has been domesticated and has four legs and a wagging tail. List C Harp Musical instrument played by pressing keys, and also has two foot pedals. Motorbike Vehicle that has two wheels, handlebars, and is propelled by pedals. Rhino Animal that has a trunk and tusks. Vase Household vessel used for measuring liquids with a spout shaped for pouring. Peacock Bird which is swift running and flightless, and buries its head in the sand. Jumper Item of clothing for a man with a collar, sleeves, and buttons down the front. Peach Fruit that is small and round, can be glazed and put on top of cakes. Spanner Tool with a squared blade, which is used for sculpting stone or wood. Rabbit Animal which squeaks and likes cheese. Frying pan Household utensil used for flattening dough. Repetition Priming Stimuli List A Tomato Item of food that is made into ketchup or can be pureed. Sheep Animal that produces wool from its coat. Toaster Household electrical appliance that browns bread by heat. Belt Item of material worn around the waist, with a buckle. Beetle Insect with hard protective casing on back and is also a make of car. Saw Tool used to cut wood with a to and fro movement. List B Potato Vegetable which can be baked, mashed or roasted. Donkey Animal that gives rides to children at the seaside. Saucepan Household item which sits on the hob, has a long handle and can have a lid.
|Printer friendly Cite/link Email Feedback|
|Author:||Vitkovitch, Melanie; Rutter, Claire; Read, Alison|
|Publication:||British Journal of Psychology|
|Date:||Aug 1, 2001|
|Previous Article:||The role of time pressure on the emotional Stroop task.|
|Next Article:||Anger on and off the road.|