تشخیص شیب قشر زیرین تغییر اوج مهاری در یادگیری: پایه عصبی برای حافظه کاذب

Experience often does not produce veridical memory. Understanding false attribution of events constitutes an important problem in memory research. “Peak shift” is a well-characterized, controllable phenomenon in which human and animal subjects that receive reinforcement associated with one sensory stimulus later respond maximally to another stimulus in post-training stimulus generalization tests. Peak shift ordinarily develops in discrimination learning (reinforced CS+, unreinforced CS−) and has long been attributed to the interaction of an excitatory gradient centered on the CS+ and an inhibitory gradient centered on the CS−; the shift is away from the CS−. In contrast, we have obtained peak shifts during single tone frequency training, using stimulation of the cholinergic nucleus basalis (NB) to implant behavioral memory into the rat. As we also recorded cortical activity, we took the opportunity to investigate the possible existence of a neural frequency gradient that could account for behavioral peak shift. Behavioral frequency generalization gradients (FGGs, interruption of ongoing respiration) were determined twice before training while evoked potentials were recorded from the primary auditory cortex (A1), to obtain a baseline gradient of “habituatory” neural decrement. A post-training behavioral FGG obtained 24 h after three daily sessions of a single tone paired with NB stimulation (200 trials/day) revealed a peak shift. The peak of the FGG was at a frequency lower than the CS while the cortical inhibitory gradient was at a frequency higher than the CS frequency. Further analysis indicated that the frequency location and magnitude of the gradient could account for the behavioral peak shift. These results provide a neural basis for a systematic case of memory misattribution and may provide an animal model for the study of the neural bases of a type of “false memory”.

Learning and memory concern the acquisition and storage, respectively, of experience. Experience consists of the multiplicity of sensory events in the several sensory systems. How experiences are represented and retained in the brain constitute central problems in neuroscience. An accurate record of experience requires precision by perceptual, acquisition and storage processes. Yet even when they are all functioning in an optimal manner, the content of the resultant memory can be different from the actual experience. “Peak shift” constitutes such a case. It consists of the systematic displacement of behavioral performance from a training stimulus (e.g., tone) to another stimulus (e.g., another tonal frequency) despite the fact that only the former had been reinforced. Peak shift is found in both classical and instrumental discrimination training, across sensory modalities and dimensions, in both humans and animals (Purtle, 1973). For example, if a 1.0 kHz tone is rewarded (CS+) while a 1.2 kHz tone is not rewarded (CS−), then the peak of the post-training frequency generalization gradient will probably not be found at the CS+ frequency but rather at a lower frequency, e.g., 0.7 kHz. Note that the peak is shifted away from the CS− to a lower frequency, in this case, because the CS− is higher than the CS+. (The opposite occurs when the CS− is lower than the CS+.) In short, although discrimination training might be thought to promote accuracy of memory of the CS+, actually peak shift reveals an impairment of veridicality of the memory of an experience. Spence, 1937 and Spence, 1942 proposed that within a stimulus dimension (e.g., acoustic frequency) rewarded stimuli (S+) produce a surrounding gradient of excitation and non-rewarded stimuli (S−) produce a gradient of inhibition. He further held that the two gradients combine algebraically with the sum dictating the gradient of resultant behavior. Insofar as the sum of the gradients would cause the peak of excitation to be displaced from the CS+, away from the CS−, Spence’s theory could in principle explain peak shift (Fig. 1). Although Spence’s ideas were published about 75 years ago, his formulation remains the dominant explanation (Bouton, 2007).