Monthly
288 pp. per issue
6 x 9, illustrated
ISSN
0899-7667
E-ISSN
1530-888X
2014 Impact factor:
2.21

Neural Computation

January 1, 2001, Vol. 13, No. 1, Pages 87-111
(doi: 10.1162/089976601300014646)
© 2000 Massachusetts Institute of Technology
A Statistical Theory of Long-Term Potentiation and Depression
Article PDF (512.18 KB)
Abstract

The synaptic phenomena of long-term potentiation (LTP) and long-term depression (LTD) have been intensively studied for over twenty-five years. Although many diverse aspects of these forms of plasticity have been observed, no single theory has offered a unifying explanation for them. Here, a statistical “bin” model is proposed to account for a variety of features observed in LTP and LTD experiments performed with field potentials in mammalian cortical slices. It is hypothesized that long-term synaptic changes will be induced when statistically unlikely conjunctions of pre- and postsynaptic activity occur. This hypothesis implies that finite changes in synaptic strength will be proportional to information transmitted by conjunctions and that excitatory synapses will obey a Hebbian rule (Hebb, 1949). Using only one set of constants, the bin model offers an explanation as to why synaptic strength decreases in a decelerating manner during LTD induction (Mulkey & Malenka, 1992); why the induction protocols for LTP and LTD are asymmetric (Dudek & Bear, 1992; Mulkey & Malenka, 1992); why stimulation over a range of frequencies produces a frequency-response curve similar to that proposed by the BCM theory (Bienenstock, Cooper, & Munro, 1982; Dudek & Bear, 1992); and why this curve would shift as postsynaptic activity is changed (Kirkwood, Rioult, & Bear, 1996). In addition, the bin model offers an alternative to the BCM theory by predicting that changes in postsynaptic activity will produce vertical shifts in the curve rather than merely horizontal shifts.