The experimental design was identical to that of Lissek et al. (2008) and consisted of three phases. During the preacquisition phase, both the S+ and the S− were each presented six times without the outcome. During the following
acquisition phase, the S+ and the S− were each presented 12 times. Of these 12 presentations, the S+ was followed 8
times by the outcome (due to a programming error, this amount of reinforced trials was slightly smaller than in Lissek
et al. (2008), where the S+ was reinforced 9 times). All the other presentations of the S+ and the S− were not followed by the outcome. Hence, the S+ was 67% reinforced. During the generalization test phase, all GSs were presented 6 times. The S+ and the S− were each presented 12 times during this phase, the S+ under a 50% reinforcement schedule (“steady-statetesting”). The test phase was divided into 6 test blocks that each consisted of one presentation of all 8 GSs, two nonreinforced presentations of the S−, and one reinforced and one nonreinforced presentation of the S+. The test trials with the GSs were always nonreinforced. The two experimental groups differed only with regard to the S−, which was a ring in group INTRA and a square in group INTER.