|
Abstract: Storing memories of ongoing, everyday experiences requires a high degree of plasticity, but retaining these memories demands protection against changes induced by further activity and experience. Models in which memories are stored through switch-like transitions in synaptic efficacy are good at storing but bad at retaining memories if these transitions are likely, and they are poor at storage but good at retention if they are unlikely. We construct and study a model in which each synapse has a cascade of states with different levels of plasticity, connected by metaplastic transitions. This cascade model combines high levels of memory storage with long retention times and significantly outperforms alternative models. As a result, we suggest that memory storage requires synapses with multiple states exhibiting dynamics over a wide range of timescales, and we suggest experimental tests of this hypothesis. (Neuron. 2005; 45(4):599-611. Cascade Models of Synaptically Stored Memories. Fusi S, Drew PJ, Abbott LF ) .FIG 2: Schematic of a Cascade Model of Synaptic Plasticity: There are two levels of synaptic strength, weak (brown) and strong (turquoise), denoted by + and ?. Associated with each of these strengths is a cascade of n states (n = 5 in this example). Transitions between state i of the ± cascade and state 1 of the opposite cascade take place with probability qi (arrows pointing up and to the left or right), corresponding to conventional synaptic plasticity. Transitions with probabilities pi± link the states within the ± cascades (downward arrows), corresponding to metaplasticity. |
|
|
|
Abstract: In many neurons, synapses increase in strength as a function of distance from the soma in a manner that appears to compensate for dendritic attenuation. This phenomenon requires a cooperative interaction between local factors that control synaptic strength, such as receptor density and vesicle release probability, and global factors that affect synaptic efficacy, such as attenuation and boosting by active membrane conductances. Anti-spike-timing-dependent plasticity, in combination with nonassociative synaptic potentiation, can accomplish this feat even though it acts locally and independently at individual synapses. Analytic computations and computer simulations show that this combination of synaptic plasticity mechanisms equalizes the efficacy of synapses over an extended dendritic cable by adjusting local synaptic strengths to compensate for global attenuationperform both sensory and motor functions. (J Neurophysiol. 2004 May;91(5) Equalization of synaptic efficacy by activity- and timing-dependent synaptic plasticity. Rumsey CC, Abbott LF ) .FIG 1 Spike-timing window functions. The plotted function, F(T), determines the amount and sign of changes in synaptic strength due to pre-post spike pairs separated by a time interval T. Pre-before-post ordering corresponds to T 0 and post-before-pre to T 0. The vertical scale is unspecified in these schematic diagrams because it depends on a number of factors (see METHODS for specific values). A: spike-timing-dependent plasticity (STDP). Pre-before-post ordering leads to long-term potentiation (LTP), and postbefore-pre to long-term depression (LTD). B: anti-STDP. Pre-before-post ordering leads to LTD and post-before-pre to LTP. FIG. 2. Equalization of synaptic efficacies by anti-STDP in an equivalent cable model. In these plots, distance is measured in units of the electrotonic length constant of the cable, and each dot represents 1 synapse. A: initial, maximal excitatory synaptic conductances, normalized to their initial value and plotted as a function of the distance of the corresponding synapse from the soma. Originally all synapses have the same strength. B: equilibrium conductance values after equilibration of anti-STDP. Synaptic strengths increase as a function of the distance of the synapse from the soma. C: initial excitatory postsynaptic potential (EPSP) amplitudes. Red dots, EPSPs measured at the site of the synapse; blue dots, EPSPs measured at the soma. Amplitudes were measured by activating synapses one at a time. D: equilibrium dendritic and somatic EPSP amplitudes. The increase of dendritic EPSP size with distance (red dots) is sufficient to equalize EPSP sizes at the soma (blue dots). E: initial synaptic efficacy decreases with distance. F: after anti-STDP, synaptic efficacy is equalized, independent of synaptic location. |
|
Steps leading to the output of the field L stage of the model. In all panels, the horizontal axis represent time and the color represents the amplitude, with blue the lowest and red the highest. In A, the vertical axis is frequency and, in the other panels, it represents the preferred frequency of the corresponding STRF. A) A song spectrogram. The outputs of the field L STRF filters applied to the spectrogram. C) The STRF outputs after normalization. |
Abstract: In songbirds, nucleus HVc plays a key role in the generation of the syllable sequences that make up a song. Auditory responses of neurons in HVc are selective for single syllables and for combinations of syllables occurring in temporal sequences corresponding to those in the bird's own song. We present a model of HVc that produces syllable- and temporal-combination-selective responses on the basis of input from recorded bird songs filtered through spectral temporal receptive fields similar to those measured in field L, a primary auditory area. Normalization of the field L outputs, similar to that proposed in models of visual processing, plays an important role in the generation of syllable-selective responses in the model. For temporal-combination-selective responses, NMDA conductances provide a memory that allows inhibitory neurons to gate responses to a final syllable in a sequence on the basis of responses to earlier syllables. When the same network that produces temporal-combination-selective responses is excited by a nonspecific timing signal, it generates a similar pattern of output as it does in response to auditory song input. Thus, the same model network can perform both sensory and motor functions. (J Neurophysiol 2003 Jan 22; A Model of Song Selectivity and Sequence Generation in Area HVc of the Songbird. Drew PJ, Abbott LF)
|
!!
