Synaptic plasticity and its physiological functions

<p indent="0mm">Plasticity is adaptive neuronal changes that can occur at different compartments of a neuron, including soma, dendrite and synapses. Among them, synaptic plasticity may exert the most significant computational power because of its high dimensionality. Synaptic plasticity is activity-dependent adjustment of synaptic weights, a concept initially postulated by theoretical psychologists and later confirmed in biological experiments. It is widely recognized that synaptic plasticity is the most critical substrate for learning and memory. Each neuron typically receives one thousand to ten thousand synaptic inputs. How these synapses are modified specifically to generate new animal behaviors is a fundamental question in neuroscience. Despite extensive research over the past few decades, it remains unclear how specific forms of synaptic plasticity participate in the animal’s learning process, primarily due to the scarcity of <italic>in-vivo</italic> studies. In this review, I examined the research on different forms of synaptic plasticity, including mechanisms both with and without supervised signals. Unsupervised plasticity can be further divided into correlation- and timing-based plasticity. Hebbian plasticity, the earliest proposed plasticity rule, lies at the center of correlation-based plasticity. Later research has revealed that not only the correlation of spikes in pre- and post-synaptic neurons, but also their temporal sequence, is critical for plasticity induction, as demonstrated in spike timing-dependent plasticity (STDP). Supervised plasticity encompasses diverse mechanisms. Canonical theories have emphasized the importance of neuromodulators, such as dopamine and serotonin, as supervised signals. Recent studies, however, have revealed a significant role of glutamatergic inputs in the instruction of plasticity, exemplified by behavioral time-scale synaptic plasticity (BTSP). In BTSP, plateau potential, a special form of regenerative dendritic calcium spike, can induce strikingly rapid plasticity in co-active synapses. Plateau potentials in CA1 pyramidal cells are typically evoked by strong inputs from the entorhinal cortex onto CA1 distal dendrites and can convert a silent cell into a responsive cell in a single trial. Compared to previously reported plasticity mechanisms such as STDP, the plasticity kernel time window in BTSP extends to the scale of seconds. Dendritic calcium spikes have been observed in other cell types, such as layer 5 pyramidal cells in the neocortex. However, whether these calcium spikes can trigger BTSP in neocortical cells remains to be studied. In addition, this review introduces recent technological breakthroughs in the manipulation of plasticity <italic>in vivo</italic>. Two different optogenetics tools have been developed to transiently block long-term potentiation (LTP) without affecting neuronal activities, which facilitates the investigation of the causal relationship between LTP and animal behaviors. Finally, I propose that the advancement in understanding synaptic plasticity would greatly benefit from the development of novel imaging and computational tools. Computational models may play a pivotal role in linking specific plasticity algorithms to experimentally observed neural activity changes in behaving animals.