A 2 X 2 factorial experiment was conducted in a licking-suppression situation to test if a rat could count the number of shocks given in a 5·min session under signaled and unsignaled shock conditions. Groups F received three .7-sec grid shocks per session throughout 80 sessions, whereas Groups V received, on any day, one, two, three, four, or five shocks, with a mean of three shocks. The rats' counting ability was assessed in terms of the post-third-shock acceleration of licking. The results of this measure were compared between Groups F and Groups V on test days in which both groups received three shocks with the identical shock sequence. There was no evidence that rats could count under either signaled or unsignaled shock conditions. The basal rate of licking was less in groups run under the unsignaled shock condition than under the signaled shock condition. The effect of fixed/variable shock frequency upon basal rate of licking was not significant. The results are discussed with reference to the optimal shock density view of Davis and Memmott (1982).Rats in an aversive situation are known to use various cues as predictors of danger and safety. Predictors may be such salient cues as tones and lights, as well as relatively subtle ones such as temporal cues and frequency cues. In a conditioned suppression situation, the rat's baseline behavior is suppressed in the presence of cues that predict danger, and the behavior recovers in the absence of dangerous cues or in the presence of safe cues. When, in an aversive situation, such cues are absent or when rats are unable to use these cues even when such cues are physically present, the baseline behavior is "chronically" suppressed.In their recent paper, Imada and Nageishi (1982) gave considerable evidence supporting the view that rats can use external cues and temporal cues. As to the rat's ability to use frequency cues, however, the results were rather inconclusive. The present paper, then, addressed the problem of the rat's ability to use a frequency cue as a predictor of safety. Davis and Memmott (1982) recently made an excellent, extensive review of counting behavior in animals and concluded that successful demonstrations of learning to count in animals were most likely to occur under relatively extreme experimental conditions in which alternative predictors of food or safety were unavailable. Experimental studies on counting behavior in rats by Davis and his collaborators formally started with the study by Davis, Memmott, and Hurwitz (1975). They found that when three unsignaled shocks were invariably given superimposed This research was supported in part by a grant in aid of scientific research awarded to Hiroshi lmada for the years 1982-1983 by the Murao Foundation, Kobe, Japan. The authors' mailing address is: Department of Psychology, Kwansei Gakuin University, Nishinorniya, Hyogo 662, Japan. 396 upon a 30-min session of barpressing, rats came to behave as though they had learned, "If three shocks, then no more shocks." More specifically, barpressing was ...