Time-dependent LTD as a stability factor

Citation
P. Munro et G. Hernandez, Time-dependent LTD as a stability factor, NEUROCOMPUT, 38, 2001, pp. 775-780
Citations number
12
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
NEUROCOMPUTING
ISSN journal
09252312 → ACNP
Volume
38
Year of publication
2001
Pages
775 - 780
Database
ISI
SICI code
0925-2312(200106)38:<775:TLAASF>2.0.ZU;2-8
Abstract
Several laboratories in studies of various brain areas have reported a depe ndence of long-term potentiation (LTP) and long-term depression (LTD) on th e relative timing of presynaptic and postsynaptic activities. The observati on that LTP occurs when presynaptic activity precedes postsynaptic activity is consistent with Hebbian ideas regarding prediction. However, the comput ational role of LTD under the opposite condition (postsynaptic precedes pre synaptic) does not find an easy explanation. Here, we propose that this for m of LTD stabilizes learning of temporal associations. This conjecture is d emonstrated in two tasks, and compared with another possible stabilizing me chanism, weight decay. (C) 2001 Elsevier Science B.V. All rights reserved.