کدگذاری محرک های حیوانات: قهقرا، حافظه اپیزودیک و برنامه ریزی های آینده

When animals code stimuli for later retrieval they can either code them in terms of the stimulus presented (as a retrospective memory) or in terms of the response or outcome anticipated (as a prospective memory). Although retrospective memory is typically assumed (as in the form of a memory trace), evidence of prospective coding has been found when response intentions and outcomes are particularly salient. At a more abstract level is the question of whether animals are able figuratively to travel back in time to recover memories of past events (episodic memory) and forward in time to predict future events (future planning). Although what would constitute adequate evidence of episodic memory and future planning is controversial, preliminary evidence suggests that animals may be capable of both forms of subjective time travel.

When animals use past experiences to make current decisions, the memory is sometimes referred to as retrospective. If they use those past experiences to anticipate future responses, it is often called prospective memory. A distinction between retrospective and prospective memory can be made in principle by asking, when animals are required to delay a response, what do they remember during the retention interval. Do they remember the characteristics of the preceding event, retrospectively, or do they remember what response they will make at the end of the retention interval, prospectively? This question, first clearly articulated by Honig and Thompson (1982), has important implications for the nature of coding processes in animal memory. Specifically, if animals have the capacity to encode prospectively it suggests that they have the ability to represent future events. Retrospective versus prospective memory Pavlovian conditioning The simplest case of the anticipation of future events is Pavlovian conditioning. In Pavlovian conditioning, when a neutral stimulus (a conditioned stimulus, CS) is paired with a biologically important event (an unconditioned stimulus, US) that produces an unconditioned response (UR), one often sees evidence of the anticipation of the US in the form of a response (a conditioned response, CR) made to the CS. Although Pavlovian conditioning has often been thought of as a process involving simple reflexive responses, and thus having few cognitive implications, recent evidence suggests that the nature of what is learned can be quite complex (see e.g., Miller & Matzel, 1988). Nevertheless, the response measure used in Pavlovian conditioning is typically a reflexive response similar to that produced by the US. Better evidence for a prospective memory process would be obtained if a future event could serve as a cue for an instrumental choice response. Simple versus conditional delayed discriminations Honig and Wasserman (1981) tested pigeons for prospective memory using an instrumental go/no-go procedure. They trained one group of pigeons on a delayed conditional discrimination (or successive matching task) in which an initial stimulus (or sample) indicated which terminal stimulus (or comparison) would be followed by food. Thus, a red or green sample stimulus indicated whether pecking the single stimulus that followed (vertical or horizontal lines) would be reinforced. If the sample was red, then pecking the vertical lines but not the horizontal lines would be reinforced. If the sample was green, then pecking the horizontal lines but not the vertical lines would be reinforced. A second group of pigeons was trained on a delayed simple discrimination involving the same stimuli but the contingencies of reinforcement were different. In the delayed simple discrimination, the same red or green sample stimulus was presented at the start of each trial but it indicated whether reinforcement would follow or not (independent of the stimulus that followed the color). Thus, for example a red sample indicated that pecking either line orientation would be reinforced, whereas a green sample indicated that pecking neither line orientation would be reinforced. Honig and Wasserman (1981) argued that once the tasks had been well acquired and a delay was introduced between the hue sample and the line-orientation comparison stimulus, if the pigeons were remembering the samples during the delay (i.e., they were coding the samples retrospectively) the retention functions for pigeons performing the delayed conditional discrimination and those performing the delayed simple discrimination should look quite similar. That is, memory for the samples should decline similarly with increasing delay. On the other hand, if the pigeons were remembering a response intention—what they were going to do at the end of the retention interval (i.e., they were coding the comparisons prospectively) then the task for the pigeons performing the delayed simple discrimination should be easier (a decision to peck or not peck) than for pigeons performing the delayed conditional discrimination (e.g., a decision to peck if the comparison stimulus was vertical lines but to refrain from pecking if it was horizontal lines). Honig and Wasserman (1981) found that the retention functions for the pigeons that had acquired the delayed simple discrimination were quite shallow, whereas those for the pigeons that had acquired the delayed conditional discrimination declined more rapidly with increasing delay (see Fig. 1), and they concluded that the pigeons were prospectively coding their response intentions.

With regard to the question of retrospective versus prospective memory, it appears that animals are capable of both. Under many conditions animals represent stimuli already experienced. However, it appears that they can also anticipate the occurrence not only of biologically important events but also of neutral stimuli like lights and tones. The best evidence for prospective memory comes from the use of differential outcomes as cues for comparison choice in conditional discriminations. It is particularly impressive when those differential outcomes do not have differential hedonic value because it suggests that animals can represent those stimuli and are not just responding to the differential hedonic states that those outcomes elicit. It should be noted, however, that the evidence for flexible retrospective and prospective representation of alternatives already selected and those not yet selected in the radial maze task may rely on unconfirmed assumptions that may be problematic. The question of episodic memory and future planning is a bit more complicated because the concepts are more abstract. The fact that we do not have an adequate definition of episodic memory makes it difficult to show evidence for it in animals. Clearly, the what, where, and when criterion is inadequate. The idea of unexpectedly asking a question about a past event has greater potential but it relies on the assumption that there is no expectation that a question would be asked about the past event. Furthermore, the evidence for episodic memory in animals obtained in this way (primarily with pigeons) involves events that occurred in the very recent past (no more than a few seconds before). Whether such effects can be found with longer delays remains to be seen. Suddendorf and Corballis (2007) have questioned the adaptive value of episodic memory and have concluded that it provides the information needed for future planning. The ability to plan or simulate the future would have adaptive value because in many cases it would avoid trial and error learning. Although evidence for future planning in animals is quite modest, especially if it is required that the animals not be reinforced specifically for that behavior (as in a delay of reinforcement procedure), fortunately, recent evidence with scrub jays (Correia et al., 2007 and Raby et al., 2007) suggests that they have some ability to plan for the future in the absence of current need. Episodic memory and future planning in animals are exciting areas of research that challenge the comparative psychologist to discover effective procedures to assess the ability of animals to use them. What once seemed to be insurmountably difficult problems to study may be limited only by our inability to discover clever ways to take the perspective of the animals we are studying.