Effects of distractors on target identification and feature integration theory.

Effects of distractors on target identification and feature integration theory

Department of Psychology, University of Warwick

Abstract

The effect of distractors on target identification was studied using 118 participants, from Warwick University. The experiment was based on work done by Treisman and Gelade (1980), who developed the feature integration theory. Each participant completed a visual search computer program consisting of three conditions. As well as these conditions as a factor there was also the varying display sizes such as 4, 8 and 16, and with the target being either present or absent in the trials. The mean response time and percentage errors were recorded. As predicted the display size had little effect on reaction time. The response rate was faster in the conjunction search condition when targets were present as not all of the distractors would have to be searched. In the single feature colour search the absence/presence of the target had no effect on response rate but it did in the single feature shape condition when the target was absent.

Introduction

Something that everyone does naturally when they are looking for an object is look out for pieces of relevant visual information. Yet somethings are easy to find whereas other items are not. This could be due to the presence of irrelevant objects or distractor items. An example of this could be looking for a child amongst a group of school children all wearing the same coloured uniform. In this case the distractor item could be the colour aspect which makes it harder to immediately spot that one child than if all the children were dressed differently. Thus it is apparent that a target with certain, distinct features, is easier to find than a target that blends in or has similarities with the distractor items surrounding it.

The aim of this practical is to investigate some of the factors that affect how easily people can find a specified target amongst varying distractors. The investigation was based on the earlier work of Treisman and Gelade (1980) who proposed a model they called the Feature Integration Theory of attention that suggests that features are registered early, automatically, and in parallel across the visual field, while objects are identified separately and only at a later stage, which requires focused attention. Henceforth this experiment focuses on the reaction times of people in visual search exercises where distractor items and conditions such as orientation, size and colour are varied. The investigation aims to test the assumption that the visual scene is initially coded along a number of separable dimensions such as the above mentioned.

This experiment uses a visual search paradigm method as a basis. During visual search tasks efficiency is measured by the reaction time taken to detect a target item against a varying number of distracters.

Treisman and Gelade (1990, cited in Treisman and Gelade, 1990) found search times increased linearly with display size in conjunction conditions, but display size had little effect for the single feature condition.

Method

Participants

One hundred and eighteen male and female, undergraduate students from University of Warwick aged between 80-20 voluntarily participated in the study. Half the pupils were asked to

...

This is a preview of the whole essay

Method

Participants

One hundred and eighteen male and female, undergraduate students from University of Warwick aged between 80-20 voluntarily participated in the study. Half the pupils were asked to

Materials

The experiment used a program called the visual search presentation software that was run on separate computers. The participants were all given instructions on a handout called ‘Using the stimulus presentation program’. The computer program presented the participants with firstly a short demo block of trials, followed by a practice block of trials and then finally the actual experiment. The program then allowed the participants to run an analysis program to analyse the results of their experiment. The full experiment consisted of 120 trials.

Design

The design of the experiment was a within subjects design as all one hundred and eighteen participants carried out the same demo trial, practice trial and most importantly full experiment. There were three independent variables that were tested during the experiment. The first independent variable was what the target looked like in terms of condition. For example there were three types of features for the object, such as did the participant have to look out for a single feature task where the target was defined by a unique colour (SF1), or secondly there was a single feature task where the target was defined by a unique shape (SF2) and thirdly there was a conjunction search task where the target was only defined by a combination of both shape and colour. Within each target condition there were another two independent variables. The second was the display size, which varied in a choice of 4, 8 and 16. Finally the third independent variable was whether the target was present or absent. Thus this was a 3 x 2 x 3 fully within subjects design. The two dependent variables were the Mean Response Time (MRT) and the Mean Percentage of Error.

Procedure

The one hundred and eighteen participants were placed in front of a computer with the program ready to run. The brief instructed them to double click on the visual search icon to begin. To start with the participants carried out the demo, which presented them with a short block of trials and then also the practice block. This enable the participants to familiarise themselves with the keys and targets. The participants were reminded they could withdraw from the experiment at any time if they wished.

The full experiment was then run via the program by each participant. The participants had to select which of the block of the three possible conditions they wanted to run and then carry them out. For example a block of single feature colour stimuli which they had to look for in each separate frame that appeared and then hit either Z if they saw it or M if they couldn’t. Thus their reaction times for spotting the distractor item or spotting its absence was recorded. But to ensure that the experiment was fair and not influenced by the buttons they had to press, half the group had to press Z as if it equalled absent and the M button as if it was present and the other half of the group had to do the reverse. This way the experiment would not hopefully be influenced by people being weaker in one hand than the other. For example faster at pressing Z for present because that’s done using their right hand.

Altogether the participants ran three full experiments firstly a block of the single feature colour targets, then a block of single feature shape targets and finally the conjunction target. The single feature colour meant the participants had to spot a target that was a blue item among green distractors. The single feature shape was a vertical bar, target among horizontal distractors. The conjunction condition was a blue vertical bar, target among blue horizontal bars and green vertical distractors.

Within each condition there were also the display sizes varying from sizes 4, 8 or 16 and the target being either present or absent. The first trial was started by pressing space button. A trial consisted of a blank screen that appeared for 500 ms and was then followed by a fixation cross for another 500 ms and then finally a display containing 4, 8 or 16 items. The target was only present on half the trials and participants had to search the display and then decide if the target was present or absent. The subjects were asked to respond as quickly as possible but not at the cost of accuracy. At the beginning of each block the participant was shown an example of what the target they were supposed to look out for was.

They were reminded of their right to withdraw and asked if they had any queries.

After the experiments were carried out an analysis program was used to

Results

For a table of results see the appendix.

Graph 1 Mean reaction times for the single feature search (colour)

Graph 2 Mean reaction times for single feature search (shape)

Graph 3 Mean reaction times for conjunction search (two features)

Discussion

The majority of the results appear to support the predictions, as the display size has little or no effect on mean response time in the single feature conditions, but in the conjunction conditions response time increases with display size. In the single feature colour condition there is little difference in the response rates in both the colour present and colour absent conditions, suggesting the feature intergration theory is correct in all of these areas. The conjunction search the response time in the target absent condition is double that in the target present condition. For the single feature shape search when there is no target present the reaction time increases with display size. This could possibly be down to personal differences between people, as the standard deviation in this condition was large and increased as display size increased. Despite this, most of the results can be explained in terms of the feature intergration theory.

One explanation for the reaction time being far quicker in the target present condition than the target absent condition in conjunction search is that the searches are “serial and self-terminating”, (Bruce and Humphreys, 1995) that is, participants search through all items until they find, or do not find the target

The response times increased with display size in the conjunction search, this is as according to the feature intergration theory (Treisman and Gelade, 1980) objects are registered automatically and in parallel. However, during the second processing stage, attention is required to identify objects.

The increase in mean percentage error in the on conjunction task compared to the single feature tasks could be due to attention being directed towards other objects (the distractors). This would mean the features of the target object are “free floating” so that there is an incorrect combination of these features and the target is wrongly identified. This is known as “illusory conjunction”, reported by previous researchers such as Efron and Yund (1974, cited in Treisman and Gelade, 1990).

A factor that may have affected the results is the fact that several participants were in the same room at the same time although on separate computers carrying out the visual search task. This may have caused distractions or slower reaction times if people paused to talk or ask questions amongst themselves although instructed otherwise.

References

Treisman and Gelade (1980). A feature integration theory of attention. Cognitive Psychology (journal), 12, 97-136.

Treisman, A. & Sato S. (1990). Conjunction search revisited. Journal of Experimental Psychology: Human Perception and Performance, 16, 459-478