The ‘Stroop effect’ illustrated the costs of automatic processes (Stroop, 1935, as cited in Edgar, 2007, p.21). Participants responded to stimuli that varied on two levels, words which were the names of colours but displayed in different coloured inks. The task was to name the ink colour; however, participants were unable to suppress the effect of word meanings which led to increased processing time. This suggested the automaticity of reading impeded the less practiced, controlled process of colour naming. However, if the directly implied word meanings were altered to have a more indirect association, would automatic processing interference still occur?
The present experiment explored this idea by altering the previous ‘Stroop effect’ design. The research hypothesis was that processing time of a conflicting colour related word task will be slower than a neutral word task. This is a one tailed hypothesis. The null hypothesis was there will be no difference in processing time of the tasks.
Method
Design
A within participant design was employed. The independent variable was conflicting word meanings. This consisted of two conditions; a ‘conflicting’ word condition in which participants identified the ink colour of words with colour associated meanings but displayed in opposing colours. The ‘non-conflicting’ word condition was displayed in the same colour sequence but word meanings had no colour association. The dependent variable was the time taken to complete the task. This was measured in seconds by the researcher and rounded to the nearest second. Confounding variables were reduced by screening participants for dyslexia, visual impairment, reading ability and colour blindness. Additionally, participants were kept naive to the experimental hypothesis. Moreover, both conditions were identical in delivering the same instructions, matching words for length, containing the same colour sequence and layout. Furthermore, counterbalancing was employed with half the group performing the conflicting condition first whereas the other half began with the neutral condition.
Participants
In all twenty participants were recruited for the procedure from colleagues at the Open University or their family and friends. The group consisted of fourteen females and six males aged between eighteen and sixty nine.
Materials
Two different lists of 30 words, the visual stimuli for each task, were produced in the font Ariel Bold at a size of 36pt and printed on individual sheets of matt white A4 paper. The first list (Condition 1) contained six conflicting words; sky, plum, lemon, grass, carrot and blood. The second list (Condition 2) contained six neutral words; sty, plan, ledge, grade, career and blame. All words were matched for length and repeated five times in their respective lists whereby the order was randomly assembled into two columns of fifteen words. Each word was printed in one of six colours; red, orange, blue, purple, green and yellow. These ink colours were randomly distributed between the words but the same sequence was replicated over both lists (Appendix 1). Additionally, a stopwatch was used to time participants’ performance and a data recording table (Appendix 2) produced to record participants’ performance. Furthermore, participants completed a consent form (Appendix 3) and were read a standard set of instructions (Appendix 4).
Procedure
Prior to testing, each participant was approached and asked if they would be prepared to take part in a cognitive psychology experiment on colour recognition which would last around 10 minutes. BPS Ethical guidelines on ‘informed consent’ were met by those agreeing to take part signing a consent form. Though participants were not informed of the experimental hypothesis (which may have duly impacted results) they were informed of the experiments extremely low levels of physical and psychological harm and what the experiment entailed thus meeting BPS ethics of ‘responsibility’. It was explained that ‘privacy and confidentiality’ would be maintained and recorded data being assigned only with a participant number. After communicating participants right of withdrawal at any time the experiment began (BPS, 2009, pp.2-24).
In the experiment itself, participants’ basic demographic information was recorded by the researcher. They were shown colour test sheet to avoid any confusion of colours (Appendix 5). The same instructions (Appendix 4) were then read verbatim to each participant, this clarified the task and participants were asked if the instructions were understood, if not the instructions were repeated. Once understood, the task proceeded whereby one of the stimulus sheets was presented face down in front of the participant. Odd numbered participants received ‘Condition 1’ sheets whereas even numbered participants received ‘Condition 2’ sheets. Participants were asked to turn the paper over and they began to identify ink colours. At this point, the researcher started a stopwatch to measure participants’ performance. On completion of the task, the stopwatch was stopped and the time recorded. Following this, participants were again read the instructions and asked if these were clear. Once understood the second ‘Condition 1 or 2’ stimulus sheet, as appropriate, was placed face down in front of participants. The previous procedure was repeated exactly with participants’ performance recorded. Following the end of the task, participants were fully debriefed as to the psychological nature of the ‘Stroop effect’ phenomenon. They were offered the chance to ask questions after which, the experiment was bought to a close.
Results
The experimental hypothesis was ‘processing time of a conflicting colour related word task will be slower than a neutral word task’. Each of the twenty participants’ response times for the two conditions is shown in the completed data set (Appendix 6).
Table 1: Mean and standard deviation for colour identification response time in the two experimental conditions.
As ‘Table 1’ shows, the mean response time of performing the conflicting word task took 2.35 seconds longer than the neutral word task. This time difference between the two conditions central tendency is illustrated in ‘Figure 1’.
Figure 1: Mean colour identification response times for the two experimental conditions.
Additionally, mean response times from counterbalancing are shown in ‘Table 2’
Table 2: Mean response times due to counterbalancing
Attempting condition 1 first took 0.7 seconds longer than if it was performing second. Similarly, performing condition 2 first took 0.4 seconds longer than if it was attempted second. These correlations in mean performance times show practice effects occurred and are illustrated in ‘Figure 2’. Therefore, the experimental counterbalancing measures were justified.
Figure 2: Response times of condition attempted first
Furthermore, a paired-samples t-test was conducted on the data. This revealed there was a statistically significant difference between participants’ mean response times with (t (19) = 3.837, p < 0.05, one-tailed, d = 0.858). The result of a large effect size and low probability of occurring by chance allowed rejection of the null hypothesis thus accepted the experimental hypothesis.
Discussion
The present experiments results showed altering distractor words to have an indirect, conflicting meaning with ink colour did produce increased response times. Response times in the conflicting condition where participants received incongruent, semantically related distractor words to displayed ink colours were statistically significantly different to the non-conflicting word condition. This result suggests automatic processing of distractor words and controlled processing of colour naming simultaneously interfered with each other.
Stroop found similar results where participants found it hard not to respond with the colour the word described rather than ink colour (as cited in Edgar, 2007, p21). This result supported a two process theory of attention which Schneider and Shiffrin distinguished as a division of attention into automatic and controlled processes (as cited in Edgar, 2007, pp.20-1). The present results further support this theory by suggesting the quicker, automatic reading of words impedes the slower conscious process of colour naming which requires attentional resources.
In this way, the findings appear consistent with Kahneman’s capacity theory which is not restricted to dealing with information from a single modality (as cited in Edgar, 2007, p.11). As such, attention being allocated to the distractor words was a momentary intention and a natural response habit deemed more relevant to the task. Therefore, this suggests automaticity operates on a continuum with controlled processes.
To this effect, the results dispute Broadbent’s ‘bottleneck’ theory (as cited in Edgar, 2007, pp.17-9). If a single filter attends to one message at a time, unattended distractor words wouldn’t have been processed thus similar performance times seen. Similarly, Treisman’s ‘attenuation’ theory is challenged (as cited in Edgar, 2007, p.19) Though unattended distractor words were not completely blocked from further processing, if early selection had occurred it would have been due to the tasks high ‘perceptual load’ which in fact was the opposite (Lavie, 1995, as cited in Edgar, 2007, pp.19-20). Therefore, the results appear to support Deutsch and Deutsch’s ‘late selection’ theory (as cited in Edgar, 2007, p.19). From Lavie’s perspective, the ‘perceptual load’ involved with the task was low thus distractor words were perceived and caused interference because of the availability of resources to process them (as cited in Edgar, 2007, pp.19-20). As such, this suggests parallel processing occurred as reading is a highly skilled process, however, without resource limitation contradicts Kahneman’s proposals.
The present study could be criticised for being artificial and lack ecological validity as it was performed in isolation from other cognitive, motivational and situational factors. Participant numbers were limited; recruiting further sample groups would provide a fairer representation. Furthermore, though none of the participants made any mistakes on the tasks, some hesitated and the experiment failed to accommodate these facts. This could provide further insight to attentional processes.
In conclusion, altering the response stimuli to a less directly implied association with a colour, automatic semantic processing interference still occurred and concurs with Stroop’s original results. The results therefore offer support for two process theories of attention. However, the relationship between bottom-up and top-down processing remains unclear. This requires further investigation with experimental enhancement.
References
Edgar, G. (2007). Perception and attention. In D. Miell, A. Phoenix, A. & Thomas K. (Eds.), Mapping Psychology (2nd ed., pp. 1-50). Milton Keynes: The Open University.
Miell, D., Phoenix, A., & Thomas, K. (Eds). (2007). Mapping Psychology (2nd ed.). Milton Keynes: The Open University.
The British Psychological Society. (2009). Code of ethics and conduct. Leicester: The British Psychological Society. Retrieved January 17, 2011, from http://www.bps.org.uk/document-download-area/document-download$.cfm?file_uuid=E6917759-9799-434A-F313-9C35698E1864&ext=pdf
