Mazur J E, Coe D
J Exp Anal Behav. 1987 May;47(3):287-97. doi: 10.1901/jeab.1987.47-287.
This experiment tested for transitivity in pigeons' choices between variable-time (VT) and fixed-time (FT) schedules. In a discrete-trials procedure, a subject chose between two alternatives by making a single key peck. Each choice was between a "standard alternative," which was the same schedule throughout a condition, and an "adjusting alternative," in which the delay to reinforcement was systematically increased or decreased many times a session. These adjustments enabled an approximate indifference point to be identified--the value of the adjusting delay at which the subject chose each alternative about equally often. Each test of transitivity involved four conditions. In one, the standard alternative was a variable-time schedule with a 2-s reinforcer, and the adjusting alternative also delivered a 2-s reinforcer. A second condition was similar except that the adjusting alternative delivered a 5-s reinforcer. The indifference point from each of these conditions was then converted to a fixed-time schedule for subsequent comparisons in the third and fourth conditions, respectively. Each of these last two conditions compared one of the fixed-time schedules (based upon the previous conditions and including their different reinforcer durations) with an adjusting schedule that delivered the alternative reinforcer duration, to determine whether the obtained indifference points would be those predicted from the prior alternative-duration comparisons with the VT schedule. There was little evidence for intransitivity of choice: Averaged across subjects and replications, the obtained indifference points deviated from perfect transitivity by less than 8%, and these deviations were not statistically significant. These results contrast with those of Navarick and Fantino (1972), who found frequent violations of transitivity between periodic and aperiodic schedules using a concurrent-chains procedure with variable-interval schedules in the initial links.
本实验测试了鸽子在可变时距（VT）和固定时距（FT）两种时间表之间选择的传递性。在离散试验程序中，受试者通过单次啄键在两个选项之间做出选择。每次选择都是在一个“标准选项”和一个“调整选项”之间进行，“标准选项”在整个实验条件下的时间表相同，而“调整选项”中强化延迟在每次实验中会多次系统地增加或减少。这些调整使得能够确定一个近似的无差异点——即调整延迟的值，在这个值下受试者选择每个选项的频率大致相同。每次传递性测试涉及四个条件。在一个条件下，标准选项是一个2秒强化物的可变时距时间表，调整选项也提供2秒强化物。第二个条件类似，只是调整选项提供5秒强化物。然后将这两个条件下的无差异点分别转换为固定时距时间表，用于后续第三和第四个条件的比较。最后这两个条件中的每一个都将其中一个固定时距时间表（基于先前条件，包括不同的强化物持续时间）与一个提供另一种强化物持续时间的调整时间表进行比较，以确定获得的无差异点是否是根据先前与可变时距时间表的替代持续时间比较所预测的那些。几乎没有证据表明存在选择的非传递性：在受试者和重复实验中进行平均，获得的无差异点与完美传递性的偏差小于8%，并且这些偏差没有统计学意义。这些结果与纳瓦里克和法蒂诺（1972年）的结果形成对比，他们在初始链接中使用带有可变间隔时间表的并发链程序，发现周期性和非周期性时间表之间频繁违反传递性。
