Computed Tomography Scans of Alcoholics : Cerebral Atrophy ?

time each day, in closed, sound-insulated chambers, with continuous white masking noise. There were thus no distractions to interfere with the performance of the task. Another feature that may have influenced the result is that the rat's first response in the session was usually reinforced. Contrast effect may have contributed to the widespread impression that BSR results in rapid extinction. After watching a rat respond at 100 per minute an experimenter may regard a rate of 5 per minute as extinction, which it clearly is not. Suboptimal electrode placement or parameters of stimulation may account for other failures to achieve good BSR with lean schedules. Sidman et al. (4), for example, used septal and caudate electrodes. Our results do not necessarily imply that chaining of responses would not produce further improvement. As pointed out by Cantor, however, most of the reported chaining experiments are confounded by presentation of multiple BSR's at the end of each chain. Cantor's own experiment, which he now claims to be an example of chaining, uses only single BSR's, but, in common with others he has presented no data from unchained control experiments. It may be that such a control is impossible. On any intermittent schedule of reinforcement the unreinforced responses are links in a chain leading to the reinforced response. This applies even to the data of Sidman et al. (4), the primary source of the belief that performance on lean schedules is impossible with BSR. The most obvious difference between food reinforcement and BSR is not in degree of chaining but in the fact that fooddeprived animals have a mechanism for motivating food-getting behavior while there is no such clearly recognizable motivation for BSR. In a deprived animal expectation of food is readily aroused because it is already being facilitated by hunger. At the beginning of a session, expectation of BSR should be more comparable to expectation of food in a satiated animal. Satiated animals will initiate responding for food (4) but quickly discover that food is no longer reinforcing and stop responding. An animal initiating responses for BSR, on the other hand, will have its expectation of reward confirmed and will continue to respond. In fact, it will probably do so with increased vigor because the association of BSR with the situational cues benefits from a recency effect. RICHARD J. BENINGER Department ofPsychology, University ofBritish Columbia, Vancouver, Canada FRANCE BELLISLE Laboratoire Jacques Le Magnen, College de France, Paris, France PETER M. MILNER Department ofPsychology, McGill University, Montreal, Canada