Dopamine dependent setting of a circadian oscillator underlying the memory for time of day.

Animals learn and remember the time of day that significant conditions occur, and anticipate recurrence at 24-h intervals, even after only one exposure to the condition. On several place-conditioning tasks, animals show context avoidance or preference only near the time of day of the experience. The memory for time of day is registered by a circadian oscillator that is set at the time of the training. We show that manipulations of dopamine (DA) neurotransmission can set a time memory in place preference and avoidance tasks, indicating that time of day is part of the context that is learned. Single injections of the DA agonist, d-amphetamine sulfate given without further exposure to the conditioning apparatus, can reset the timing of anticipatory behavior evoked by previously acquired place-event associations. The data support a model for time memory in which DA signaling sets the phase of a circadian oscillator, which returns to the same state at regular 24-h intervals. The data also raise the possibility that some apparent impairments of memory formation or retention could reflect post-experience resetting of the optimal retrieval time rather than impairment of memory or retrieval per se.

Memory for time of day (time memory) is encoded by a circadian oscillator and is distinct from other context memories.

We report that the neural representation of the time of day (time memory) in golden hamsters involves the setting of a 24-h oscillator that is functionally and anatomically distinct from the circadian clock in the suprachiasmatic nucleus (SCN), but is entrained by the SCN acting as a weak zeitgeber. In hamsters, peak conditioned place avoidance (CPA) was expressed only near the time of day of the learning experience (± 2 h) for the first days after conditioning. On a 14:10 light:dark cycle, with conditioning at the end of the light period (zeitgeber time 11 [ZT11]), CPA behavior, including time of day memory, was retained for more than 18 d. With conditioning in the early day (zeitgeber time 03 [ZT03]), CPA was completely lost after 5 d but reemerged after an additional 6 d, with the peak avoidance time shifted to ZT11. When the entraining light cycle was shifted immediately following learning at either ZT11 or ZT03, with no additional experience in the training apparatus, peak CPA 18 d later was always found at ZT11 on the shifted light cycles. When conditioned at ZT03, then placed into constant dark for 18 cycles, the peak shifted to subjective circadian time 11 (CT11). In all experiments, the peak CPA time was set initially to the time of experience, and was reset subsequently to the end of the subjective day, without memory loss for other context associations. In the absence of an SCN, peak avoidance was not reset. Therefore, time memory is distinct from other context memories, and involves the setting of a non-SCN circadian oscillator. We suggest that circadian oscillators underlying time memory work in concert with the SCN to enable anticipation of critical conditions according to both immediate- and long-term probabilities of where and when important conditions could be encountered again.