Cellular Basis of Memory

Cellular Basis of Memory

On the basis of lesion analytical studies, the hippocampus and related structures have been implicated as a site for the con-solidation of memory. The question naturally arises out of the properties of the neurons in this area that might permit the as-sociation of information. Neuroanatomically (Figure 15.2), the hippocampal formation consists of the fields of Ammon’s horn (regions CA₁, CA₂, CA₃), the dentate gyrus and the subiculum. Afferent pathways from the entorhinal cortex project to the hip-pocampus via the perforant pathway and synapse on the granular cells in the dentate gyrus. The entorhinal cortex itself receives cortical inputs from polysensory associational regions in the frontal, temporal and parietal lobes. Within the hippocampal formation, the granule cells in region CA₃ also project to the CA₁ region through the fimbria fornix, which also projects to the subiculum. The subiculum is the major efferent pathway, project-ing to a number of cortical regions but also projecting back to the entorhinal cortex, completing the loop.

For learning and memory to occur, there must be plastic changes such that the structure and functional characteristics of nerve cells and their interconnections are altered. Much of the re-search into the processes of synaptic plasticity underlying learn-ing and memory has been conducted in invertebrates (Kandel et al., 1983; Kandel, 1991) through examination of reflexive learning processes (habituation, sensitization and classical conditioning).

Regarding the mammalian brain, Bliss and Lomo (1973) were the first to demonstrate that repeated stimulation of the

afferent pathways to the dentate granule cells of the hippocampus of the rabbit produced an excitatory potential in the postsynaptic hippocampal neurons lasting for hours. Recording in intact ani-mals has shown potentials that lasted for days and weeks. They termed this increased facilitation as a result of repeated stimula-tion long-term potentiation (LTP).

With respect to area CA₁ of the hippocampus, studies have shown that LTP occurs only when a number of input pathways have been stimulated. This is known as the criterion of coopera-tivity. When distinct weak and strong excitatory inputs impinge on a pyramidal nerve cell, the weak input becomes potentiated through association with a strong input. This is known as the cri-terion of associativity. Finally, the criterion of specificity refers to

the fact that strong repeated stimulation in one synaptic pathway is specific only to that stimulated pathway. Unstimulated syn-apses on the same cell do not demonstrate LTP.

The events surrounding LTP are schematically described here (Figure 15.3); the reader is referred to a detailed explana-tion of the cellular and molecular aspects of LTP (Kandel, 1991Shepherd, 1994). Studies of the CA₁ region of the hippocampus reveal that LTP is mediated by the neurotransmitter glutamate.

LTP has been stressed as a cellular mechanism of infor-mational connectivity in the hippocampus. Brief mention should also be made of long-term depression (LTD). LTD is the opposite of LTP and refers to “use-dependent long-lasting decreases in synaptic strength” (Linden and Connor, 1995). LTD may have a number of advantages and work in parallel with LTP with respect to memory functioning. LTD may help reset synapses that have been potentiated by LTP, to prevent saturation. It may serve as a cellular mechanism of forgetting (Tsumoto, 1993) and, finally, may also form an active inhibitory system to attenuate signals from adjacent potentiated synapses. The specific role of LTD in the coordination of memory is still unclear. For a detailed de-scription of the potential cellular and molecular mechanisms in-volved in LTD, the reader is referred to other sources (Linden and Connor, 1995; Linden, 1994). The phenomena of LTP and LTD provide an example in which neural cytoarchitecture and the underlying cellular and molecular levels may actually conform to the principles of association and connectivity. Computational models have been developed to further our understanding of the manner in which learning and memory emerge from the proper-ties of synaptic plasticity embodied in the circuitry of the hip-pocampus (Churchland and Sejnowski, 1992; Traub and Miles, 1991). The attraction of computational models is that they may clarify the link between aspects of LTP and LTD within local networks of associated neurons at the cellular level and the events at the behavioral level.