The derivation of the conclusion ''Anna is bigger than Mary'' from the
premises ''Anna is bigger than Paul'' and ''Mary is smaller than Paul
'' is considered an instance of transitive deduction. For a non-verbal
presentation, the premise statements were here transformed into a mul
tiple operant discrimination task. Adult subjects were trained with ov
erlapping pairs of a six-member stimulus series (A+B-, A+C-, C+D-, D+E
-, E+F-; +: choice rewarded, -: choice penalized). A computer game-typ
e presentation that hid the actual problem structure from the subjects
was employed. The effects of varying the presentation style of the ta
sk on the objective performance and the structure awareness of subject
s were investigated. A first experiment used random polygons as stimul
i and the relations between them were only signalled by the above rein
forcement allocations. In a second experiment the stimuli were cartoon
figures additionally involved in a dominance hierarchy that was sugge
sted graphically. A third experiment used named items that were relate
d through visible size differences in addition to the reinforcement al
locations but was otherwise like an experiment using an abstract forma
t reported by Werner et al. (1992). In all experiments a similar propo
rtion of subjects responded transitively when subsequently tested with
the pairs ED, BE and CE by preferentially choosing stimulus B or C. E
ach subject subsequently filled in a questionnaire, completed a stimul
us ordering exercise, and was interviewed to find out whether they wer
e explicitly aware of the stimulus hierarchy underlying each of the ta
sks. Although the proportion of subjects revealing an explicit transit
ive responding increased together with the concreteness of the stimuli
and their relations across the experiments, the objective performance
in terms of choice accuracy did not vary. The accuracy performance on
tests could be accurately simulated with a modification of a simple c
onditioning model. It is concluded that an implicit mode of processing
may underlie many instances of transitive responding in humans even w
hen explicit task understanding is reported.