Learning-Induced Physiological Memory in Adult Primary Auditory Cortex: Receptive Field Plasticity, Model, and Mechanisms

Abstract

It is well established that the functional organization of adult sensory cortices, including the auditory cortex, can be modified by deafferentation, sensory deprivation, or selective sensory stimulation. This paper reviews evidence establishing that the adult primary auditory cortex develops physiological plasticity during learning. Determination of frequency receptive fields before and at various times following aversive classical conditioning and instrumental avoidance learning in the guinea pig reveals increased neuronal responses to the pure tone frequency used as a conditioned stimulus (CS). In contrast, responses to the pretraining best frequency and other non-CS frequencies are decreased. These opposite changes are often sufficient to shift cellular tuning toward or even to the frequency of the CS. Learning-induced receptive field (RF) plasticity (i) is associative (requires pairing tone and shock), (ii) highly specific to the CS frequency (e.g., limited to this frequency ± a small fraction of an octave), (iii) discriminative (specific increased response to a reinforced CS+ frequency but decreased response to a nonreinforced CS– frequency), (iv) develops extremely rapidly (within 5 trials, the fewest trials tested), and (v) is retained indefinitely (tested to 8 weeks). Moreover, RF plasticity is robust and not due to arousal, but can be expressed in the deeply anesthetized subject. Because learning- induced RF plasticity has the major characteristics of associative memory, it is therefore referred to as ‘physiological memory.’ We developed a model of RF plasticity based on convergence in the auditory cortex of nucleus basalis cholinergic effects acting at muscarinic receptors, with lemniscal and nonlemniscal frequency information from the ventral and magnocellular divisions of the medial geniculate nucleus, respectively. In the model, the specificity of RF plasticity is dependent on Hebbian rules of covariance. This aspect was confirmed in vivo using microstimulation techniques. Further, the model predicts that pairing a tone with activation of the nucleus basalis is sufficient to induce RF plasticity similar to that obtained in behavioral learning. This prediction has been confirmed. Additional tests of the model are described. RF plasticity is thought to translate the acquired significance of sound into an increased frequency representation of behaviorally important stimuli.