Abstract
In two experiments, pairs of Garner’s classical 5-dot patterns were presented with an interstimulus interval of 500 ms in a same–different task in which a physical sameness criterion was used: Rotated or reflected versions of the same pattern were rated as different. Patterns varied in “goodness” according to Garner’s equivalence set size measure. Both first and second pattern goodness affected reaction time and accuracy. This result and fits of models to reaction time data indicate that equivalence set representations are used in the task, as in a related categorical matching task in previous studies. Two effects were observed that contrast with the categorical matching task: One is a conflict between the need to respond different to patterns that are categorically equivalent under the equivalence set representation; the other is that extra time is needed for rechecking of the representation if pattern structures are hard to distinguish. In combination with previous studies, the present results show that even though the processes differ, the same representational mechanism is used across tasks.
Acknowledgments
This work was supported by Grant La 1281/2-1 from the Deutsche Forschungsgemeinschaft (DFG) to Thomas Lachmann. Many thanks are due to Robert Proctor, Lester E. Krueger, Beverly Roskos-Ewoldson, Peter van der Helm, and John Flowers for critical and very helpful remarks on earlier drafts of this paper.
Notes
One reviewer (John Flowers) suggested an alternative explanation of the results of Experiment 2. He argued that an overt strategy shift may have occurred in some participants. This is because of the reduction in the number of patterns and the greater similarity among them as compared to Experiment 1. The shift would occur for the present physical sameness instructions, as opposed to the categorical sameness instructions of our previous studies. The new strategy would involve feature scanning. When a particular figure shows up as first pattern, they prepare during the ISI to scan for the presence of a contraindicating feature in the second pattern. Such a process would be likely to overrule the ESS effect in different responses. In fact, if it were a learned optional strategy, more likely to be activated by ESS 8 patterns, it could indeed produce a slope reversal.