Negative Priming: The effect of inhibitory mechanisms on the probe of a pair of trials in a Stroop style ink identification task.Abstract
The investigation was based on the work of Dalrymple-Alford and Budayr (1966), who investigated the phenomenon of negative priming in relation to the Stroop task. In the original experiment by Dalrymple-Alford and Budayr (1966), it was discovered that if in a trial, the ink colour was the same as the word on the previous trial; subjects were slower to respond. This effect has been termed negative priming. The aim of this experiment was to partly replicate the work of Dalrymple-Alford and Budayr (1966), and to further investigate the phenomena of negative priming. The experimenter hypothesised that in an ink colour identification task, when the target in the probe trial matched the distractor in the prime, then reaction times would be significantly slower in comparison to conditions where the prime and probe were unrelated. To test the hypothesis, the researcher created four conditions; congruent, neutral, ignored repetition and attended repetition. The condition of interest was ignored repetition. Participants reaction times were recorded for the primes and probes of each condition. The effect of condition was shown to be significant using a two way repeated measures ANOVA [F(3,57) = 13.09; p = 0.001]. The significance of the results means the hypothesis was accepted, and it was concluded that negative priming is prominent in conditions where the target in the prime becomes the distractor in the probe, supporting the work of Dalrymple-Alford and Budayr (1966).
Attention is a vital and complex function of cognition. One of the earliest definitions of attention came from James (1890), who defined it as “the taking possession by the mind, in clear and vivid form, of one out of what seem several simultaneously possible objects or trains of thought…It implies withdrawal from some things in order to deal effectively with other.” This early definition from James (1890) highlighted the issue of the selective nature of attention. This feature of attention is essential for organisms to be able to be successful in a search for a target; to select and process only the information they need. It is therefore vital that during this search there are certain mechanisms that suppress distracting information and prevent the return of attention to previously attended objects or events. The mechanism responsible for this important feature is inhibition- the suppression of unwanted or distracting information to ensure movement of attention to novel locations.
The role of inhibition has been theorised through a variety of concepts. One such concept is Inhibition of Return (IOR). IOR was proposed as an inhibitory mechanism, which reduces the prominence of the previously inspected item in a scene. IOR was first observed by Posner and Cohen (1984) in their simple cuing experiment and refers to the relative suppression of stimuli (object and events) that had recently been the focus of attention. This inhibition of return effect is thought to make visual search more efficient as it ensures that previously examined objects are not searched again, thus facilitating the search for the target (Wright & Richard, 1996).
Further evidence of inhibitory mechanisms in attention comes from the visual marking mechanism; proposed by Watson and Humphreys (1997) as a goal-directed process that enhances visual search through the inhibition of ‘old’ objects. When new objects are added to a visual scene, they take priority during search, because old objects are ‘marked’ for non search. Also, the discovery of the ‘attentional blink’ provides some clear evidence that in tasks using Rapid Serial Visual Presentation (RSVP), a method of displaying information very briefly in sequential order, perception of a target presented 200-500ms after the first target is impaired (Raymond, 1992). This attentional blink occurs because of interference caused by the presentation of stimuli after the target but before the target-identification process is complete, causing the temporary suppression of inhibitory mechanisms. In other words, inhibition of distracting stimuli does not occur, causing a failure in identification.
Mechanisms such as IOR and visual marking are evidence for inhibitory mechanisms in selective attention, and the attentional blink demonstrates just how important these mechanisms are. This process of inhibition however, is not without consequences. It has been discovered that after a stimulus has been ‘ignored’, processing of that ignored stimulus shortly afterwards is impaired. This effect has been termed negative priming (Tipper, 1985). In recent years, numerous studies have looked at negative priming as evidence of an inhibitory component within selective attention.
An important study which was vital in the discovery of negative priming is the Stroop task (Stroop, 1935). The standard Stroop colour-word test involved participants being required to name the ink colour of a printed word. When the word was incongruent with the colour ink- such as the word ‘red’ written in green ink- then interference occurred, resulting in slower response times and more errors in comparison to control conditions. In congruent conditions, where the colour written matched the colour of ink it was written in, reaction times were faster. The interference observed in this study can be attributed to automaticity as reading is an automatic process. According to Shiffrin and Schneider’s (1977) model of automaticity, automatic processing makes no demands on attentional resources, has no capacity limitations, and is unavoidable. Automatic processing thus provides a liable explanation of why the Stroop effect occurs, as when seeing a word we unavoidably read it, causing a delay in the process of naming the ink colour of the word.
Whilst investigating the effect of stimulus sequencing on Stroop interference, Dalrymple-Alford and Budayr (1966), came across what is now known as negative priming. What they found was that there was a greater delay and an increased error rate when an item appeared in the colour ink which was required to be ignored in the previous stimulus. Similar findings come from Tipper (1985), who presented participants with overlapping line drawings, in either red or green. The participants were required to identify only the red items in each set of stimuli. When the ignored drawing (green) became the required response (red) in the next set of trials, response times slowed. This suggests it is harder to identify and selectively attend to what was previously rejected.
A key question in regards to selective attention and negative priming is at which point of sensory processing can incoming signals first be selected or rejected by attention- does this happen early in the process or late? Early selection models, such as Broadbent’s (1958) filter theory, argue that as sensory processes are limited, they require attention to initially select the stimuli that are required for further processing and discarded irrelevant stimuli. Therefore, attentional selection should occur early; implying a ‘bottleneck’ in the brain protecting processing systems from being overloaded by irrelevant information.
The late selection models (Deutsch & Deutsch 1936) however, claim that all stimuli, both attended and unattended, can be processed automatically in parallel- thus without a need for early selection. Therefore, selection should occur late, after the semantic analysis of the stimuli. Negative priming has generally been interpreted as evidence for late selection as the phenomenon shows that distracting/ irrelevant stimuli are in fact processed at the same time as the attended stimuli, hence the interference that occurs.
Negative Priming is clearly a well studied phenomenon, and there have been numerous variations on the original experiment by Dalrymple Alford and Budayr (1966). The explanations behind the effect have generally focused on the effect being caused by increased interference due to the suppression of the word during naming of the ink colour- resulting in temporary unavailability of that response (MacLeod & MacDonald, 2000). The majority of evidence supports the idea that if a probe in a pair of stimuli has the same target as the prime, then reaction times will be slowed for that probe; suggesting that internal representations of the ignored object may become associated with inhibition during selection. Therefore this experiment hypothesises that, in concordance with the previous evidence, in an ink colour identification task, the probe in the ignored repetition condition will take significantly longer to identify than the prime, in comparison to other conditions.
The design was repeated measures with 2 within factors; condition with 4 levels (Congruent, Neutral, Ignored Repetition and Attended Repetition) and pairing with 2 levels (prime and probe). The experiment was a part replication of the work of Dalrymple-Alford and Budayr (1966), as an investigation into negative priming.
The experiment consisted of 4 conditions. Condition 1 was ‘Congruent’, where the target and distractor matched in both prime and probe, for example blue in blue ink followed by red in red ink. Condition 2 was ‘Neutral’, where the normal Stroop style format was used and the prime and probe bore no intentional resemblance to each other; for example blue in red ink followed by yellow in green ink. Condition 3 was ‘Ignored repetition’. This condition was where negative priming was presumed to take place, as the distractor in the prime became the target in the probe, for example, blue in yellow ink followed by red in blue ink. The final condition, condition 4, was ‘Attended repetition’, where the target was repeated in the probe, for example blue in red ink followed by green in red ink.
For each condition, there were 30 pairs of trials (120 pairs in total, 240 individual trials). Within each pair was a prime (1) and a probe (2) The trials were split into two identical blocks. To control for order effects, the conditions were randomised such that no condition/ pair was presented in succession. This resulted in 15 pairs of each condition per block. A total of 240 responses (reaction times, in milliseconds) were collected for each participant.
The sample selected was a group of 20 undergraduate students at the University of Lincoln, with a mean age of 21.35 years, and a standard deviation of 6.51. This target population was relevant because it was the most easily accessible group of people of similar age and status. Participants were selected by opportunity sampling. This method was used because it is a quick, practical and efficient way of generating data through using participants available and willing at the time of the experiment.
In order to carry out the experiment certain materials were necessary. The researcher used a Dell Optiplex 745 computer with a monitor size 15inches, 150HP. Also used was a button box (Credus Corporations) and voice recorder (TTC Quality Electronics). The 6 colours used were randomly selected from a bag of various coloured cards. The chosen colours were then created from a standard Microsoft window’s palette. These were;
Blue (red: 0, green: 0, blue: 225), Green (red: 0, green: 225, blue: 0), Red (red: 225, green: 0, blue: 0), Yellow (red: 255, green: 255, blue: 0), Pink (red: 225, green: 0, blue: 225), Black (red: 0, green: 0, blue: 0). All colour words were presented in Aerial font, size 58, bold.
In addition to the colour words presented, there was also a welcome message (Arial font, size 48, bold, in Black ink), and a fixation cross (Arial font, size 58, bold, in Black ink).
Further necessary materials included a checklist for Type I and Type II errors.
The participants were approached and asked if they would like to take part in the experiment. If they agreed they were taken to a quiet area chosen for the experiment to take part in. Then the researcher explained to the participant what they would need to do, and gave them a set of standardised instructions (appendix 1). The participants were then asked to read and sign the consent form (appendix 2) if they agreed to take part. Following this, the participants were seated in front of the computer screen and shown how to hold the microphone. They were then told there would be an initial practice run of the experiment, and asked to begin when they were ready. Following the practice run, the participant was once again asked if they were happy to continue with the experiment. If they agreed, they were instructed to begin when they were ready.
During the experiment, two researchers were present at all times. The researchers each had a list of the order of trials and correct responses, as they were pseudo-randomised. One researcher marked type I errors on one sheet, and the other marked type II errors on another. Block one consisted of a series of 60 trials followed by a 30second break before the remaining 60 trials in that block. The experiment began with a welcome message which instructed the participant to press the left key on the button box when they were ready to start. After they had pressed this, a fixation cross was presented on the screen for 1500ms, followed by a blank which lasted 1000ms. Each trial was presented for 1500ms, trials were presented in pairs according to condition. Between each pair was a blank of 1000ms.
After the first block of trials, the experiment closed, and one researcher started block two, which was identical to block one. Once again any and all errors were recorded. After the completion of this second final block, the experiment automatically closed. The participant was then thanked for their cooperation and given a debrief form to read (appendix 3) they were also encouraged to ask any questions, and assured that their results would remain private and anonymous.
A number of ethical issues were identified in the experiment in line with British Psychology Society (BPS) guidelines. A consent form was given to participants which explained what the experiment was researching into, what they had to do during the testing and it also requested the participants age and gender. The form explained that any participant with aversion to flashing lights or rapidly presented stimuli should not continue on with the experiment, and asked participants to report if they had any back problems. Participants also had the right to withdraw themselves and their results from the experiment at any time, and this was stated in both the consent form (appendix 2) and debrief (appendix 3). After the participants had taken part in the experiment, the experimenter explained what they were investigating and the implications to the research, and answered any questions asked. It was the experimenter’s responsibility to make sure that participants left in the same psychological state that they started the experiment with. Participants were informed that their identity would be kept anonymous and that their results would be treated in confidence and destroyed after the experiment. To ensure protection of participants, no physical or mental harm came to them while taking part in the experiment as the consent form included a brief health check to eliminate those individuals who may be at slight risk from participating in the experiment. The room was an empty, calm setting, in order to minimise any stress to the participant, and to avoid any eye strain, a break was given, splitting the trials into two blocks. No deception took place in this experiment.
An ethical approval form was completed by researchers prior to the experiment (appendix 4).
The results were recorded and analysed for each condition in the experiment- 1 (Congruent), 2 (Neutral), 3 (Ignored Repetition) and 4 (Attended Repetition). Any errors, either cognitive (type I), or human/computer (type II), were excluded from the data. Both prime and probe trails were removed regardless of where the error occurred. Error analysis will be discussed later.
A table to show a comparison of the mean and standard deviation of the difference between reaction times of prime and probe per condition
3 (Ignored Rep)
4 (Attended Rep)
A table to show a comparison of mean & standard deviation for reaction times of prime and probe per condition
1 (Congruent ) Prime
1 (Congruent ) Probe
2 (Neutral) Prime
2 (Neutral) Probe
3 (Ignored Rep) Prime
3 (Ignored Rep) Probe
4 (Attended Rep) Prime
4 (Attended Rep) Probe
See appendix 5 for full SPSS data.
The mean difference between prime and probe for condition 3 (Ignored Repetition) was -69.35, which was significantly greater than for any of the other conditions (21.25 for Congruent; 12.95 for Neutral, and 35.45 for Attended Repetition). It also shows that the condition with the smallest difference in reaction time between prime and probe was condition 2 (Neutral).
2 shows that for condition 3 (Ignored repetition) the mean reaction time for the prime (782.7) was smaller than the mean reaction time for the probe (852.0). This stands out when compared to all of the other conditions, where the mean reaction time for the prime was greater than for the probe. This suggests that for conditions 1 (Congruent), 2 (Neutral) and 4 (Attended Rep), the probe generated a quicker response than the prime, yet for condition 3 this effect was reversed and the probe generated a slower response.
To further analyse the data, a Two-way Repeated Measures ANOVA was carried out to analyse the reaction times and look at any effect between conditions. The results of the ANOVA shows that the main effect of Condition was significant [F(3,57) = 13.09; p = 0.001]. The following bar chart ( 3) presents a visual representation of this significance and shows the variation between conditions:
The second ANOVA was concerned with the difference in reaction times between prime and probe. The ANOVA showed that the main effect of Pair is not significant [F(1,19) = 0.001; p = 0.996], suggesting that the pairing did not significantly affect reaction times. Although the effect was not found to be significant, the plot below ( 4) clearly shows that condition 3 (Ignored Repetition) was the only condition where response time was slower in the probe than in the prime:
Thirdly, the interaction effect between Condition and Pair was analysed. This was found to be significant [F(3,57) = 6.6; p = 0.001]. As the interaction effect between ‘Pair and ‘Condition’ was significant, a post-hoc Bonferroni was carried out to find where the significances lay. The Bonferroni showed significant differences between the following conditions;
(1) Congruent and (2) Neutral (p= 0.001)
(1) Congruent and (3) Ignored Repetition (p= 0.002)
(2) Neutral and (4) Attended Repetition (p= 0.001)
(3) Ignored Repetition and (4) Attended Repetition (p= 0.014)
Errors were recorded per type I and II for each condition. The table below ( 5) shows the number of errors of each type that occurred in each condition.
A table of sums of errors per condition and error type
Error Type I
Error Type II
3 (Ignored Rep)
4 (Attended Rep)
A table to show the mean rank of errors per condition
3 (Ignored Rep)
4 (Attended Rep)
The condition with the lowest number of errors was condition 1 (Congruent), with a mean of 2.15. The condition with the highest number of errors condition 3 (Ignored Repetition), with a mean of 2.93.
A Friedman’s test was used to analyse the errors and look for any significances in their distribution. Application of Friedman’s test showed that there were no significances in the distribution of errors over the four conditions; X2=5.71; df = 3; p = 0.127.
The results obtained show that the mean difference between prime and probe for condition 3 (Ignored repetition) was -69.35, which was notably greater than for any of the other conditions (21.25, 12.95 and 35.45). This suggests that something different is happening in this condition, as the difference is not only a lot greater but also in the opposite direction. The plot ( 4) shows a visual representation of this effect. From this it is possible to infer that in the Ignored repetition condition, negative priming did occur as the probe took longer to respond to than the prime in comparison with all other conditions. After carrying out a two way repeated measures ANOVA, it becomes clear that this is in fact the case. The results of the ANOVA showed that the main effect of Condition was significant [F(3,57) = 13.09; p = 0.001], and that the interaction effect between Condition and Pair was also significant [F(3,57) = 6.6; p = 0.001]. The ANOVA concerning the difference in reaction times between prime and probe showed that the main effect of Pair was not significant [F(1,19) = 0.001; p = 0.996].
Analysis of errors found them not to be significant; however the mean ranks showed that there were more errors in the ignored repetition condition (Mean rank 2.93). This is consistent with previous research; that in the ignored repetition condition, more interference occurs causing slower response times and more mistakes to be made.
These findings mean that the hypothesis can be accepted: in an ink colour identification task, when the target in the probe trial matches the distractor in the prime, then reaction times will be significantly slower in comparison to conditions where the prime and probe are unrelated. Thus the experiment supports and confirms the previous research such as that of Dalrymple-Alford and Budayr (1966).
The negative priming effect observed in this experiment can be explained as an inhibitory mechanism of attention. The differences in reaction times between conditions infer that for condition 3 (Ignored Repetition), at the point of the probe something different happened in than in the other conditions. In line with previous research, we can assume that due to the suppression of the word in the prime trial, when that colour word then becomes the ink colour in the probe trial, then there is a problem with retrieving that response as it had just been suppressed.
One limitation of this experiment was the methodology. The design involved a set of two blocks in a Super Lab program, each containing 15 pairs of each condition, in a randomised order. Between each pair of trials was a blank screen presented for 1000msc. This quick succession of pairs means it may not have been obvious for the participants that the stimuli were in fact presented in pairs. This therefore may be able to explain why the probe condition 1 (Congruent) was fastest; when it was expected that condition 4 (Attended Repetition) would be. To overcome this limitation, future experiments could use separate blocks for each condition- thus making it more obvious that the trials were in certain pairs.
In addition to the above adjustment, it would also be interesting to consider individual differences in a future extension of this experiment. There has been numerous past studies that suggest for certain individuals, the effect of negative priming is actually less robust. An example of this is Schizophrenics, who seem less able at inhibition- hence are less susceptible to negative priming (Beech et al 1989). A future investigation could build on the evidence of individual differences playing an important role in the effect of negative priming, and possibly look into more general differences such as cultural background or occupation. For example, it would be interesting to look for any differences in the effect of negative priming between people in creative careers- such as artists, compared with those in writing careers such as journalists. Would someone who is used to looking at words be more prone to negative priming than someone who would be more interested in the colour and form of the word?
To summarise, this experiment has shown clear negative priming, consistent with the majority of existing studies, thus supporting the notion of inhibitory processes in attention.
Beech, A., Powell, T., McWilliam, J., & Claridge, G. (1989). Evidence of reduced cognitive inhibition” in schizophrenia. British Journal of Clinical Psychology, 28, 109-116.
Broadbent, D. E. (1958). Perception and communication. New York: Pergamon.
Deutsch, J. A., & Deutsch, D. (1963). Attention: some theoretical considerations. Psychological Review, 70, 80-90.
Dalrymple-Alford, E.C.,& Budayr, B. (1966). Examination of some aspects of the Stroop color-word test. Perceptual and Motor Skills, 23, 1211-1214.
James, W. (1890). The Principles of Psychology. New York: Henry Holt.
MacLeod, C. M., & Masson, M. E. J. (2000). Repetition priming in speeded word reading: Contributions of perceptual and conceptual processing episodes. Journal of Memory and Language, 42, 208-228.
Pashler, H. (1998). The psychology of attention. Cambridge, MA: MIT Press.
Posner,M.I., & Cohen, Y. (1984) . Components of visual orienting. In H. Bouma & D.G. Bouwhuis (Eds.), Attention and performance X: Control of language processes. Hove: Lawrence Erlbaum Associates Ltd.
Raymond J.E., Shapiro K.L., Arnell K.M. (1992). “Temporary suppression of visual processing in an RSVP task: an attentional blink?”. Journal of experimental psychology. Human perception and performance, 18, 849-60.
Shiffrin, R.M., & Schneider, W. (1977). Controlled and automatic information processing: II. Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127-190.
Stroop, J.R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-662.
Tipper, S.P. (1985). The negative priming effect: Inhibitory priming with to be ignored objects. The Quarterly Journal of Experimental Psychology, 37A, 571-590.
Tipper, SP (2001) Does negative priming reflect inhibitory mechanisms? A review and integration of conflicting views. Quarterly Journal of Experimental Psychology 54A: 321-343.
Watson, D. G., & Humphreys, G. W. (1997). Visual marking: Prioritizing selection for new objects by top-down attention inhibition of old objects. Psychological Review, 104, 90-122.
Wright, R.D & Richard, C.M. (1996) Inhibition-of-return at multiple locations in visual space. Canadian Journal of Experimental Psychology, 50, 324-327.