A Test of Sensory Pre-Conditioning in Pigeons

Abstract

Sixteen pigeons were used in an experiment designed to show the sensory preconditioning effect as originally reported by Brogden. An experimental group received 200 simultaneous presentations of a buzzer paired with a light stimulus. They were then trained to respond to one of these stimuli alone and tested for response to the other. A control group received the same treatment except in the initial stage when 200 presentations of the test stimulus only were given. The results provided no evidence that the pairing of stimuli affected behaviour during the critical test.

Both groups responded to the test stimulus to a similar degree in this experiment, whereas in Brogden's original study with dogs experimental animals responded significantly more frequently than did their controls. The discrepancy in results can be attributed to the use of different control procedures in the two experiments. Brogden's controls were not exposed to presentations of the test stimulus before training. Differences in familiarity with this stimulus may have produced the differences between the behaviour of his control and experimental animals. This interpretation is supported by the result of a preliminary experiment with pigeons in which Brogden's control procedure was used and his original results confirmed.

With human subjects there is some evidence that pairing procedures may result in enhanced generalization, alterations in sensory thresholds or hallucinations. However, these effects are little understood and difficult to predict. Although no functional relationships like those found in ordinary conditioning have yet been shown to apply, the terms “sensory conditioning” and “sensory pre-conditioning” have been widely used and the data have been quoted in support of theories of learning that require the setting up of direct sensory-sensory relationships.

Until there is unequivocal evidence of pairing effects with animals or fuller knowledge of the occurrence of such effects with human subjects, it is considered unadvisable to link them even by name with basic learning mechanisms such as conditioning.