THE EFFECTS OF A CANT-ANSWER RESPONSE OPTION AND INSTRUCTIONS ON STIMULUS EQUIVALENCE

Two interlocking conditional discriminations were established in a pap er-and-pencil match-to-sample format by written rule-like instructions to 104 college students-specifically, ''When A is the sample, circle 1'' (not 2); ''when B is the sample, circle 2'' (not 1); when 1 is the sample, circle X'' (not Y); ''when 2 is the sample, circle Y'' (not X ). Some subjects always found a CAN'T-ANSWER response option among the comparison stimuli in the match-to-sample format; other subjects neve r did. All subjects had the same discriminations established repeatedl y under different instructional conditions: Baseline: no instructions beyond those establishing the format; then restrictive instructions, c autioning all subjects not to go beyond the information presented in t heir rule-like instructions; and finally nonrestrictive instructions, urging all subjects to look for relations implicit in their rule-like instructions. Performance of these discriminations was probed in the s ame format to reveal the extent to which they functioned as equivalenc e relations; these probes were intermixed with the ongoing performance of the instructed discriminations. The display of symmetry, transitiv ity, and symmetric transitivity, definitive of equivalence relations, was seen at best in about 70% of the subjects. Its emergence was great ly reduced by the presence of the frequently chosen CAN'T-ANSWER optio n among the comparison stimuli available to one group but not to the o ther group; and it was considerably reduced in both groups by the rest rictive instructions to answer the probes in a rule-bound manner.