Discrepancies In The Content Of Temporal Memory

Individual differences in the content of temporal memory can be evaluated by the horizontal placement of psychophysical functions that relate signal duration to the probability of a response (Church and Meck, 1988; Gibbon et al., 1984). Data obtained from the PI procedure have shown that discrepancies in the content of temporal memory produce stable horizontal displacements of timing functions such that they can be centered at times that are either less than or greater than the programmed time of reinforcement (Church, 1989; Meck, 2002a, b). Typically, the average remembered time of reinforcement for a group of mature rats would be very close to the programmed time of reinforcement, with a symmetrical distribution of individual peak times centered around that time. In contrast, as rats age they demonstrate a proportional rightward shift in their timing functions, indicating that their remembered durations reflect a constant percentage overestimate of the programmed time of reinforcement (see Lustig, this volume; Meck, 2002a; Meck et al., 1986).

In rats, the observed differences in this remembered time of reinforcement have been shown to be proportional to the programmed time of reinforcement and can be related to modifications in cholinergic (Ch) function as determined by drug and lesion studies. For example, increasing the effective levels of acetylcholine in the brain by systemic administration of physostigmine has been shown to produce a maintained proportional leftward shift in timing functions, whereas decreasing the effective levels of acetylcholine by systemic administration of atropine has been shown to produce a maintained proportional rightward shift in timing functions, relating the probability of some response to signal duration (Meck, 1983, 1996; Meck and Church, 1987). The behavioral effects of selective brain damage have also indicated differential modifications in the content of temporal memory. Lesions of the fimbria-fornix or the Ch cell bodies in the medial septal area projecting to the hippocampus have been shown to produce a maintained proportional leftward shift in timing functions, whereas lesions of the frontal cortex or the Ch cell bodies in the nucleus basalis magnocellularis projecting to the frontal cortex have been shown to produce a maintained proportional rightward shift in timing functions (Hills, this volume; Meck et al., 1987; Olton et al., 1987, 1988).

This discrepancy in the content of temporal memory has been described in terms of a multiplicative translation constant that is responsible for producing scalar transforms of sensory input taken from an internal clock (e.g., Church, 1989; Gibbon et al., 1984; King et al., 2001; Meck, 1983, 1996). An example of how this might work is as follows: The number of pacemaker pulses integrated by an accumulator during the presentation of a signal can be considered to serve as a clock reading that provides a representation of the perceived duration for the current trial. When this clock reading is transferred to reference memory (presumably following feedback), the transfer may occur with some bias that we have referred to as a memory storage or a memory translation constant. In scalar timing theory (see Gibbon and Church, 1990), this has formally been referred to as the K* parameter, which is a multiplicative constant in the equations describing how the reference memory process works during encoding and retrieval when animals make a temporal discrimination. If the remembered time of reinforcement reliably differed from the obtained clock reading, then an animal would consistently expect the feedback to occur later than the programmed time if its memory storage constant was greater than 1.0. Alternatively, it would consistently expect feedback to occur before the programmed time if its memory storage constant was less than 1.0. As indicated by Church (1989, p. 64),

Without a memory storage concept, the behavior of the rats that act as if they expect an event earlier or later than it regularly occurs would seem to violate the general principle of reinforcement in which differential reinforcement of responding at the correct time would eventually lead to time estimates that are accurate in the mean.

Memory storage speed has been used as a proposed mechanism for this translation constant whereby sensory input from the internal clock is transferred to reference memory at some modifiable baud rate that ultimately influences the quantitative aspects of the represented signal duration (e.g., Church, 1989; Gibbon et al., 1984; Hinton and Meck, 1997a; Matell and Meck, 2000; Meck, 1983, 1996). The main idea here is that the content of temporal memory (i.e., the remembered duration of an event) is based on the amount of time required to transfer the number of pulses in the accumulator (i.e., this process functions as an "up counter") to reference memory (i.e., this process functions as a "down counter"). As a consequence, the transfer time (and hence the remembered duration) would be directly related to the number of pulses in the accumulator and the speed of transfer. Presumably, the nervous system can function as if it has a conversion table that relates the amount of time the memory storage network is activated during the transfer of a specific signal duration. If the speed of this memory storage process were to deviate from normal values (i.e., conditions under which the remembered duration of an event is equal to the programmed duration of an event), then it would be possible for proportionally shorter or longer values to be represented in memory. Unlike changes in clock speed, changes in memory storage speed would not be self-correcting even in the case where animals were "surprised" by the mismatch between their clock readings and a value sampled from reference memory. This is because any "updating" of memory in this case would lead to the continued distortion of stored values for as long as the modification in memory storage speed remained in effect, i.e., as long as K* was > or < 1.0 (e.g., Church and Meck, 1988; Meck, 1983, 1996, 2002a, 2002b, 2002c).

Sodium-dependent high-affinity choline uptake (SDHACU) reflects the activity of cholinergic neurons in brain regions during behavioral activation. Using this measure, Meck (2002a) attempted to establish the quantitative relation between SDHACU in the frontal cortex and hippocampus as a function of the discrepancies in the remembered times of reinforcement. SDHACU levels should be proportional to the magnitude of the discrepancy in temporal memory in those brain areas that monitor the comparison between the current clock reading and a sample of the expected time of reinforcement taken from reference memory. This discrepancy should reflect the participant's degree of surprise at the mismatch between these two variables when feedback is provided. The concepts of surprise and expectancy are major features of contemporary theories of animal learning and memory (e.g., Gibbon, 1977; Gibbon and Balsam, 1981; Kamin, 1969; Rescorla and Wagner, 1972; Schultz et al., 1997; Wagner, 1981).

Theta rhythm activity has also been investigated as a potential index of the content of temporal memory (Meck, 2002b). Atropine-sensitive theta rhythm (4 to 12 Hz) in the frontal cortex and hippocampus appears to be necessary for certain learning and memory processes (e.g., Givens and Olton, 1990). Differences in electrical patterns (e.g., amplitude and modal frequency) can be related to individual differences in the remembered times of reinforcement stored in working and reference memory. It is also possible to impair or facilitate temporal memory by driving hippocampal rhythmical slow-wave activity at frequencies that differ significantly from those of normal patterns of rhythmic excitation (Meck, 2002b). Taken together, the results of a series of experiments suggest that theta rhythm frequency in the frontal cortex serves as an indicator of the direction of the discrepancy in the content of temporal memory and that SDHACU amplitude in the frontal cortex serves as an indicator of the magnitude of the discrepancy in the content of temporal memory (Meck, 2002a, b).

0 0

Post a comment