In two experiments, pigeons pecked side keys in a discrete-trials setting in which shorter and longer runs of successive pecks on the left key before a switch to the right key occasionally produced, after a brief retention interval, a short-term memory probe for the most recent run length. In Experiment 1, a probe involved red and green side keys. A peck to a green (red) key was reinforced if the previous run length was shorter (longer). The dependent variable was the probability of a peck to the correct color. In Experiment 2, a probe involved an autoshaping procedure in which a response-noncontingent reinforcer was delivered after a 5-secpresentation of a green (red) center key if the previous run had been a shorter (longer) one. A reinforcer was not delivered when a red key followed a shorter pattern or a green key followed a longer pattern. The production of runs conformed to many previous molecular data on the way the local temporal patterning of behavior adapts to, that is, displays knowledge of, a reinforcement contingency. The probe results showed that a pigeon can report which of two run lengths it recently has emitted. Thus, a pigeon can, in a sense, describe its own adaptive behavior. Since the adaptive behavioral patterning on the center key may be said to represent knowledge, and since the probe behavior is a self-characterization or self-report by the organism about this knowledge, the probe behavior may be said to represent knowledge about knowledge, or metaknowledge. The data extend previous work on metaknowledge in the pigeon to a third type of adaptive temporal pattern of behavior, that is, run length (instead of response duration and interresponse time), and provide a second type of probe procedure, that is, autoshaping, by means of which a nonverbal organism can be asked what it knows about what it is doing to adapt to an environmental contingency.The relation between performance and self-reports of that performance is once again assuming a major theoretical role in human cognitive psychology (Brown, 1975; Gazzaniga, 1970;Jacoby & Dallas, 1981; Nisbett & Wilson, 1977). This research relies heavily on human subjects' verbal self-reports. Related experiments, however, can also be conducted with nonverbal animals (Nelson, 1974;Shimp, 1981;Ziriax & Silberberg, 1978). For example, Shimp (1981) arranged a setting in which a pigeon produced a behavior stream consisting mostly of shorter or longer interresponse times on a center key. This behavior revealed a rather precise adaptation to the reinforcement contingency for center-key pecks. Such adaptation constitutes, in a sense, a kind of knowledge on the part of the subject about what the centerkey reinforcement contingency is. In addition, symbolic matching-to-sample trials were occasionally arranged on colored side keys: a response to a particular colored side key was reinforced if the precedThis research was supported in part by NIMH Grant 16928.The author would like to thank Stephen J. Hutton for his expert help with the conduct of this research....