Q270: Experiments and Models in Cognitive Science

Feature Search Laboratory #1

For this laboratory, you should prepare a full report describing the experiment that you conducted. It will probably be around 5 pages long, and it should include the following sections:

Introduction: This should describe what the study of human pattern recognition involves, and what theories exist for how people recognize objects. Of these theories, describe the feature-based model of pattern recognition most completely. Describe Treisman and Gelade's (1980) Experiment 1, and why the results from this experiment were interpreted as providing evidence for a feature-based model. In this section, you will want to discuss the following notions: conjunctive versus disjunctive tasks, the role of attention in binding features from an object together, the slope and intercept of the best fitting line relating array size and response time, and exhaustive versus self-terminating processes.

Methods: Describe the stimuli you used as accurately and objectively as possible, such that anybody could replicate the experiment if they so desired. Describe all of the different conditions, any randomizations that were used, what the subjects' task was, who the subjects were, and how many subjects were run. Use tables here and elsewhere if they make the experiment easier to understand.

Results: Describe the outcome of the experiment, supplying statistical tests to support any claims you make. For this experiment and all others in the class, you can use a p < 0.2 criteria for significance. You should describe the major results that were relevant for the theories described in the introduction, whether or not they were statistically significant. Importantly, provide easy-to-understand graphs for all important trends that would be hard to understand in text.

For example, you should prepare a graph comparable to Treisman and Gelade's Figure 1, showing (correct) response times as a function of display size, with different lines for conjunctive-absent, conjunctive-present, disjunctive-absent, and disjunctive-present trials.

Results that you will want to include (with means for the compared conditions, and statistical tests):

Was there a main effect of task (conjunctive versus disjunctive)?

Was there a main effect of array size (1 versus 4 versus 16 versus 36 letters)?

Was there a main effect of target presence (present versus absent)?

Was there an interaction between task and array size?

Were there any significant two-way or three-way interactions involving these variables?

What is the equation for the line that best describes the relation between array size and response time, for each of the four tasks (conjunctive-absent, conjunctive-present, disjunctive-absent, and disjunctive-present)?

You can use an ANOVA for all of these questions except the last one. The last one requires linear regression.

Discussion: How do your results challenge or confirm a feature-based account of pattern recognition? Were there any anomalous results from a feature-base perspective; if so, what could explain them? In the final analysis, what do your results suggest about how people perceive, and why is this an important finding?