Four groups of Ss were tested on hidden-word tasks consisting of 12 set-inducing items followed by 13 test items. One group, the simple-task group, received 7-letter items having a 4-letter word embedded in each. The other group, the complex-task group, received 10-letter items with each again having a 4-letter word embedded in it. The “set” response consisted of choosing alternate letters that formed a word, while the direct solution consisted of simply choosing 4 consecutive letters that formed a word. Direct solutions were possible only on the 13 test items. It was found that the complex experimental group produced significantly more direct solutions (“broke” the set more often) than did the simple experimental group. Results supported a behavioral model of rigidity based on the principles of Hull-Spence behavior theory.