Visual Search


Introduction

Why is it so hard to find a needle in a haystack? Is it because the needle is small? Likely not, if the needle were red, or the needle were shaped like a ball, it would be pretty easy to find, needles get lost in haystacks because in order to be distinguished from the hay around them two features, colour and shape, must be processed at the same time, and this takes attention. This topic has been explored experimentally using the 'visual search' paradigm (e.g., Treisman and Gelade (1980)). In this paradigm participants are asked to look for a specific target among distractor stimuli. The target may differ from distractors in one or more of its features and there may be a few or a large number of distractors. Differences in response times for target detection across condition has helped psychologists better understand many aspects of visual information processing.


Summary



The above chart shows the typical results from a visual search experiment. You can see response time for the 'feature search' condition, (where the green circle was presented alongside red circles) compared to the a 'conjunction search' condition (where the green circle was presented amidst green squares and red circles). Note that in the Feature Detection condition the number of distractors does not affect target detection time. This is because colour is processed preattentively, or automatically, and we are able to group items based on colour without attending to each item individually. This effect is known as the 'pop out effect'. Colour, luminance, orientation, motion direction and velocity all produce 'pop out' in visual search experiments, suggesting that these visual features are all processed automatically and without attention.


Here are some examples of pop out, find the 'different' stimulus in the array.


The results from the conjunction search condition are much different. The response time to finding the target increases as the number of items in the array increases. Feature Integration Theory (FIT) proposed by Treisman and Gelade (1980) explains this finding by suggesting that while individual features (such as colour and shape) are processed in parallel, binding those features into a single object, such as a target that contains features present in other items in the array, requires attention. So in the conjunction search, where it is necessary to bind multiple features of 'green and circle' to determine if an item is the target, items need to be attended to one by one until the target is found. When more items are in the array the participant must, on average, pay attention to more items before discovering the target and thus response time increase with the number of items in the array.


Review Questions

Question: Consider an array where the target was the letter 'T' and the distractors were the letter 'L'. Do you think that the number of distractors would affect how fast the target was detected?


Answer: Yes. The letter T is comprised of two features, a horizontal line and a vertical line. The letter 'L' is comprised of the same features in a different configuration. In order to bind these features and determine if an item is a target items would need to be attended to sequentially and an effect of number of distractors would be observed.


Question: Suppose you are in charge of designing a sign that will be seen by all motorists. Based on your knowledge of visual search, what characteristics could you include in your sign to increase the probability that it will be noticed?


Answer: You can use any of the preattentive features to make your sign more noticeable, but because a motorist will be in motion, the best features would be luminance, colour and line orientation/shape. A bright orange triangle will pop out of the typical outdoor surroundings based on all three of these features.