This study determined whether a two-person exchange situation contained natural contingencies for trusting behavior or whether external contingencies were necessary. Pairs of college students worked matching-to-sample problems for money. On each trial there was one problem and the subjects determined which of them would solve it. Trusting behavior was defined as an increase in the number of consecutive problems each subject allowed his partner to work during sessions that also ended with an equitable distribution. Simply, trust was a temporary deviation from equity. A subject could give the problem to the other person (cooperate), or not respond and let the other person take the problem (share). Other possibilities were for both subjects to try to take the problem (complete), or for neither subject to respond and thereby let the person who worked the last problem also work the next one (passive trust). When only four lever pulls were required to distribute a problem (no external contingencies to reach either equity or trust) subjects reached equity, but only minimal trust (strict alternation of single problems) developed in 18 sessions. When 30 or 60 lever pulls were required to distribute a problem (smaller response requirement for passive trust and therefore a contingency for trust), trusting behavior developed after a few sessions (fixed ratio 30) or after several trials of the first session (fixed ratio 60) and it ordinarily expanded gradually to 10 to 15 consecutive problems through passive trust. The aversiveness of the inequity involved in trusting appears to necessitate a contingency for acquisition. Once trust develops, however, this aversiveness is reduced as subjects learn the inequity is only temporary (e.g., once trust was acquired at fixed ratio 60 it was maintained at fixed ratio 4, which would not initially produce it), and the direction of the inequity appears to become of questionable importance (e.g., being behind was alternated over rather than within sessions and usually not in a systematic manner).

Trustful behavior was defined in terms of the consecutive numbers of matching-to-sample problems worth money that each subject worked during sessions that ended in an equitable distribution. Two stages of acquisition are inherent in this definition; the first stage requires acquisition of an equitable method of distributing reinforcers (cooperation) to show that the within-session deviations (trust) from equity that develop during the second stage are temporary and are not part of an inequitable method of distributing reinforcers. Previous research has indicated that a contingency to trust is necessary to override the aversiveness of the inequity inherent in trusting and to produce consistent and maximal trust (half of the problems worked consecutively by each subject). The present experiment examined such a contingency. The trust contingency was an increased requirement for changing the direction of problem allocation. Only the subject who had been allocated a problem could change that allocation, by pulling a lever 45 or more times. On the other hand, no separate responses were required to allow the person who worked the last problem to also work the next one (passive trust). Hence, giving a problem was the only way to increase the distribution of problems to the other person and hence prevent oneself from receiving all of the reinforcers. All eight pairs of subjects cooperated from the outset. Trusting behavior developed for all four pairs exposed to the contingency to trust and expanded to maximal levels by the second session for three of the four pairs.