PL2132: Research & Statistical Methods 2 - Lab Report

Running Head: PARALLEL AND SEQUENTIAL REPRESENTATIONS AND RECOGNITION

Effects of Parallel and Sequential Representations on Short-term Recognition

Chan Yi Tsun, Cheong Meiyi, Chua Siti Ayeshah Mohd Shafiq,

Tan Jia Xin Jacinth and Yong Yeng Hong

National University of Singapore

Abstract

This study, a 2x2 mixed factorial design, examined how parallel and sequential representation of information influenced the response time in a recognition task. Twenty-one undergraduates were randomly assigned either to the parallel representation or sequential representation condition. Participants decided whether the target house was a match or mismatch from the stimulus house presented earlier and the response time for correct answers was measured. Results supported the hypothesis that parallel processing produced a shorter response time than sequential processing. It partially supported the hypothesis that the number of features only influenced the response time for sequential processing. These findings indicate how future in-depth research could be directed at using audio description instead of written description for sequential representation.

Effects of Parallel and Sequential Representations on Short-term Recognition

The comparison between parallel and sequential representation of information has been commonly cited as an evidence for using images to perform cognitive tasks, specifically the short term recognition task. In a study by Nielsen and Smith (1973), students were shown either a schematic face picture or its verbal description with varying number and sizes of features and asked to memorize them. After a varied retention interval, they were made to match against stimuli presented and their response times were measured. Results from the study revealed that “matching was relatively fast and independent of the number of relevant features” only when the participants were shown the face picture, while the response time for matching based on the verbal description was slower and “increases as a function of the number of relevant features on the list (Nielsen & Smith, 1973; cited in Reed, 2006). These findings suggest parallel representation of information as an efficient and “useful strategy for recognizing described objects” (Finke, 1985).

A number of studies have also made similar findings which concluded reaction time as “an increasing function of stimulus information” for sequential but not parallel representation (Hick, 1952; cited in Lindsay & Lindsay, 1966). However, these studies have also pointed out a common underlying limitation of their findings in that participants could have made “multiple representations of a single form” (Posner, 1970; cited in Nielsen & Smith, 1970). Presenting the sequential representation in a written description, which was what these studies did, could result in the possibility of participants processing the verbal information in a visual form as well, which would render written description an invalid measure of sequential processing.

One of the main purposes of the present study was to control for this possibility by providing an audio description instead of a written description, while attempting to replicate the results of the current literature. The method of investigation was similar to that of Nielsen and Smith (1973), except that houses with different features were used in place of faces and the number of levels of features presented was reduced from three to two for simplification of the study.

In this regard, the present study sought to assess two hypotheses. First, it was posited that the time taken to match a perceived image to a memorized image is influenced by the type of representation – sequential or parallel. Participants in the parallel processing group were expected to produce a shorter response times than those in the sequential processing group. Second, with the number of features of the house to be processed varied at two levels – three and five, this factor should only affect the response time for sequential processing, with a longer response time for the five-features condition.

Method

Participants

Twenty-one male and female undergraduates from the Psychology statistics level-two course at the National University of Singapore participated in the study during lesson time.

Materials

A total of 34 pictures of houses were drawn out using the programme, “Paint”, out of which nine had three features (door, window and chimney) and the remaining 25 had five features (door, window, chimney, fence and path). Each of these features assumed one of two sizes: large or small (see Appendix 1).

For the parallel representation condition, 18 pictures of houses were used as stimuli and targets, with six used as both. Half of these 18 were three-featured houses and the rest were five-featured houses.

For the sequential representation condition, 12 pictures of houses were used as targets and12 pre-recorded audio descriptions of houses were used as the stimuli.

The programme, “E-Prime”, was used to present the targets and stimuli to the subjects, and to record their response time.

Procedure

The participants were randomly assigned into two conditions, with 10 participants in the parallel representation condition and 11 participants in the sequential representation condition. Each group then proceeded to carry out the experiments in two separate rooms.

For the parallel representation condition, each participant went through two practice trials followed by 10 test trials. Each trial followed the following sequence: (a) a picture of a stimulus house was projected onto a screen in front of the participant for four seconds; (b) an unfilled fixation slide appeared for four seconds; and (c) a picture of a target house whose onset started a clock appeared. The participants responded by pressing either a match button (‘1’ key) or a mismatch button (‘2’ key) as quickly and accurately as possible. The sequential representation condition followed similar procedures, except that an audio description of the stimulus house was read out to the participant in part (a) of the above sequence instead. The description always followed a fixed order of door, window, chimney, fence and path.

In both conditions, the number and sizes of house features to be remembered were randomly ordered in each trial. There were five matched and five mismatched responses in the test trials. These were also randomly-ordered. The response times for all the correct answers of every participant were recorded and analysed.

Results

Response time data were subjected to a 2 x 2 analysis of variance (ANOVA) with the type of representation as a between-subjects factor and the number of features presented as a within-subjects factor.

The mean response time of participants in the different conditions were as follows: parallel processing of three features was 2132.64ms (SD = 715.05), parallel processing of five features was 2424.58ms (SD = 751.49), sequential processing of three features was 3146.39ms (SD = 1233.53) and sequential processing of five features was 3216.55ms (SD = 954.32).

Results revealed a significant main effect for the type of representation: Response time was faster for parallel processing than sequential processing, F(1, 19) = 7.765, MSE = 1.10 x 106, p < .05. However, there was no significant main effect for the number of features: F(1, 19) = .498, MSE = 6.90 x 105, p > .05. In addition, no significant interaction effect was found: F(1, 19) = .187, MSE = 6.90 x 105, p > .05.

Discussion

The results of the present experiment supported the first hypothesis that the type of representation - sequential or parallel would influence the response time needed to determine whether a perceived image matched a memorized one. Participants who were in the parallel representation condition responded significantly faster than those in the sequential representation condition. These results were consistent with findings by Nielsen and Smith (1973).

The second hypothesis, nevertheless, was only partially supported. As predicted, the number of features in the image did not affect the response time for parallel representation. This implies that a person can maintain a visual pattern of the stimulus image in the short-term memory and all the features can be compared simultaneously with the target image.

However there was no significant result for the number of features on sequential processing in the present study. This was inconsistent with the findings reported by Nielsen and Smith (1973), which demonstrated that the response time for sequential processing was dependent on the number of features in the image. Similarly, reaction time increased as a linear function of the number of items in the memory set (Sternberg, 1967; cited in Reed, 2006). The most likely explanation for this inconsistency is that there was a lack of counterbalancing for the order of the mismatched feature in the mismatch trials of the sequential representation condition. Thus, in some trials, participants of the sequential representation condition did not have to compare all the features before making a response. This led to a lack of difference in response times between the three-feature condition and five-feature condition when the mismatched feature was the first, second or third feature to be compared. Hence, an improvement to this study would be to add more test trials, so as to vary and counterbalance the order of the mismatched feature presented.

Another limitation in the present study was that only two practice trials were provided before the actual trials. Practice trials were placed to allow participants to familiarize themselves with the matching task. In the original study by Nielsen and Smith (1973), participants went through 18 practice trials. However, this present study was not able to accommodate as many practice trials as that of the original study due to time constraints. In addition, the absence of feedback on correct or wrong responses during the practice trials could also have led participants to not fully understand the experiment and hence, made more mistakes in the practice and actual trials.

This study has managed to replicate the finding that parallel processing of information is faster than sequential processing even when audio description was used as a mode of sequential representation instead of the written description used in previous studies. Hence, future in-depth studies could adopt the use of audio description so as to minimize the possibility of participants representing written description in a visual form as well.

Similarly, a visual representation of the stimulus may not necessarily result in parallel processing of the information because it is possible for participants to memorize and recall the features individually instead of treating the whole house as a visual template to be encoded (Posner, 1970; cited in Nielsen & Smith, 1970). Hence, future studies could take this into consideration and develop with ways to ensure that parallel processing is used.

In conclusion, despite the limitations of the present study, it has managed to replicate the finding that parallel processing is a more efficient way of processing information for short-term recognition than sequential processing. It has also addressed the issue of using a more valid measure of sequential processing, which future studies be directed to use.

References

Finke, R.A. (1985). Theories relating mental imagery to perception. Psychological Bulletin, 98(2), 236-259.

Lindsay, R.K & Lindsay, J.M. (1966). Reaction time and serial versus parallel information processing. Journal of Experimental Psychology, 71(2), 294-303.

Nielsen, G.D & Smith, E.E. (1970). Representations and retrieval processes in short-term memory: recognition and recall of faces. Journal of Experimental Psychology, 85(3), 397-405.

Nielsen, G.D & Smith, E.E. (1973). Imaginal and verbal representations in short-term recognition of visual forms. Journal of Experimental Psychology, 101(2), 375-378.

Reed, S.K. (2006). Cognition: theory and applications. United States of America: Thomson Wadsworth.