The significance of circadian phase for performance on a reward-based learning task in hamsters

Abstract

In humans and animal models, circadian modulation of learning has been demonstrated on numerous tests. However, it is unclear which aspects of the cognitive process are rhythmically regulated. In these experiments, we used a conditioned place preference task in hamsters to ask whether memory acquisition (hypothesis 1) or memory recall and performance (hypothesis 2) were subject to circadian modulation. In golden hamsters, access to a running wheel has been used as a reward to condition a place preference, but when given unrestricted access to a wheel, animals perform most of their spontaneous running within a few hours each day or circadian cycle. This suggested that either the perceived reward value of the wheel changes through the day or that the response to this reward is temporally restricted. Contrary to the hypotheses, we found that learning was not tied to the time of training nor to the time of testing, but rather animals showed a preference for a reward-paired context only at the circadian time that training had taken place. Timing is not an explicit discriminative cue in these experiments. Hence, the learning mechanism must be predisposed to register circadian time as an attribute during context learning.

Introduction

A growing body of evidence demonstrates that circadian timekeeping can influence the processes of learning and memory. This comes from studies of both humans and animal models (see Ref. [12]). The ability to learn, remember, and then to modify current responses according to prior experience are fundamental attributes of animal behavior. In mammals, learning involves a complex set of steps that comprise sensory processing, acquisition, memory formation, memory retention (and extinction), and memory retrieval. Recent studies have begun to demonstrate that these processes may be differentially influenced by the animal's circadian rhythms or by the time of day.

An association between rhythmicity and learning has been demonstrated in three general ways. First, situations where circadian rhythms are disturbed are invariably linked with cognitive impairment. In humans, these include the aging process, shift work, transmeridian travel (jet lag), and disease. In animal models, the circadian disruptions produced by phase shifting the light–dark cycle result in impaired performance on passive avoidance tasks [6], [10], [27]. Age-related rhythm disturbances impair performance on a conditioned place preference (CPP) task in hamsters [1], and on a test of spatial learning in rats, memory retention is selectively impaired while the initial acquisition and recall are intact [7].

Second, processes underlying cognitive performance may be modulated over the day or a circadian cycle. Numerous studies have shown that optimal times of day exist for learning different tasks in humans and in animals [11], [15], [16], [20], [21], [29], [30]. In addition, retention deficits on numerous tasks recur at periodic intervals in rodents, suggesting that memory recall is subject to temporal modulation separately from acquisition and memory formation [18], [15], [16], [17], [29].

Finally, in some species and situations, the time of day or the phase of the circadian cycle may be an attribute of the environment that is learned. The ability to learn, remember, and respond according to the specific timing of events has an enormous adaptive significance [2], [9], [12], [13], [28]. Animal's ability to learn the timing of significant events is indicated by the anticipatory behavior of various animal species that are presented with timed rewards, e.g. temporally restricted feeding schedules [3], [8], [13], [23], [26].

Taken together, these studies show that essentially all steps in the learning process may be subject to temporal regulation. The degree to which each step is affected appears to depend on the species and the learning paradigm that is used. In addition, the time of day may be a feature that is learned along with other attributes such as location, context and value of the presented stimulus [12].

In our experiments, we have begun to address the extent to which the circadian rhythms influence each of these aspects of learning. Because reward is a strong motivator of learning and performance, we have focused on reward-associated learning. Using a CPP task, we have shown previously that the ability to learn associations between reward and context is impaired in hamsters with circadian rhythm disruption [1]. In rodents, and especially in hamsters, running wheel activity is highly rewarding. However, when given free access to a running wheel in constant environmental conditions, running activity is strongly restricted to the animal's subjective night time. This suggested two alternative hypotheses: (1) that the perceived reward value of the wheel changes through the day; or (2) that the response to reward is temporally restricted. Using brain stimulation reward (BSR), Yanofski et al. [31] showed that perceived reward value does not vary with circadian time in rats. However, we have found significant effects of BSR on hamster locomotor rhythms that are not detectable in rats (S.W. Cain et al., manuscript in preparation). Therefore, there may be important species differences in the connections among circadian and brain reward mechanisms.

Based on the first hypothesis, we predicted that the strength of the CPP that is developed by using timed wheel reward will vary with the animals' spontaneous patterns of behavior and wheel use. This was complicated by the fact that hamsters may be induced to use a wheel at times when spontaneous activity is low, i.e. a novel wheel can be potentially rewarding when the home cage wheel is not. This issue was addressed in the test of the second hypothesis from which we predicted that the acquisition of a place preference could occur any time that the relative reward value of the stimulus were high, but that the expression of the preference would still depend on the animal's natural pattern of locomotor activity.

Neither of the two hypotheses was supported by the results of these experiments. Instead, CPP expression was restricted to the temporal match between training and testing times. This indicates that the internal representation of time is a significant attribute of context even when time is not a discriminative cue.