Conditioned Reinforcement: Order Out of Chaos

The empirical research on conditioned reinforcement, after nearly half a century, has failed to yield a coherent and conclusive body of literature to justify its extensive use as a genuine phenomenon and a basic construct within the field of learning. The primary reason for this dilemma is a history of the use of inadequate controls and the use of paradigms which are subject to interpretations other than that of conditioned reinforcement. In order to determine conclusively, the necessary and sufficient conditions for the establishment of a conditioned reinforcement effect, intracranial stimulation of the lateral hypothalamus was used as the primary reinforcer in a study using multiple controls to remove and assess all possible sources of confounding. Following implantantion, each subject was first placed in a two lever chamber for three consecutive days to obtain baseline levels of responding for a neutral stimulus. Three groups of 8 subjects pressed for one neutral stimulus on one lever and no stimulus on the second lever. Two groups of 8 subjects pressed for one neutral stimulus on one lever and the second neutral stimulus on the second lever.In the second or pre training phase, all subjects were placed in a single lever chamber in which lever pressing resulted in delivery of intracranial stimulation (ICS). After three 30 minute sessions on consecutive days, subjects failing to meet a 200 press per session criterion were eliminated from the study.Subjects were next given Pavlovian training with ICS in a chamber free of manipulanda. Of the groups that had received previously only one baseline neutral stimulus, one was presented with that stimulus 0.5 s prior to ICS (forward pairing). A second group received the neutral stimulus randomly relative to ICS. A third group received the neutral stimulus in an explicitly unpaired format relative to ICS. Both of the groups that received two neutral stimuli, received one stimulus 0.5 s prior to ICS. One group received the second stimulus randomly relative to ICS, and the second group received the second stimulus in an explicitly unpaired format relative to ICS.All groups received 100 randomly spaced presentations of the neutral stimulus or stimuli per day in 100 minute sessions for four consecutive days, with ICS presented 100 times on Day 1 and an average of 50 times thereafter intermittently.Subjects were tested during one 40 minute session in the original two lever chamber. Results indicated that subjects significantly increased responding for tie forwardly paired stimulus from baseline to test. Increases in responding for control stimuli from pretest to test were either non-significant or significant but weaker than increases for the forwardly paired stimulus. Results were interpreted as clearly demonstrati.ng a conditioned reinforcement effect, for which Pavlovian temporal contiguity is, empirically, the necessary and sufficient condition. The results were discussed in terms of various theories of conditioned reinforcement and a revised model was proposed, emphasizing the arousal of a hypothetical fractional neural excitatory response and its conditioning to a CS as the basis for all conditioned reinforcement effects.