TY - JOUR
T1 - Nonmonotonic Plasticity
T2 - How Memory Retrieval Drives Learning
AU - Ritvo, Victoria J.H.
AU - Turk-Browne, Nicholas B.
AU - Norman, Kenneth A.
N1 - Funding Information:
This work was supported by National Institute of Mental Health ( NIMH ) R01MH069456 to K.A.N. and N.B.T.B., and an National Science Foundation ( NSF ) Graduate Research Fellowship to V.J.H.R. We are grateful to Judy Fan, Chris Honey, Justin Hulbert, Jarrod Lewis-Peacock, Ehren Newman, Jordan Poppenk, Randy O’Reilly, Andrew Saxe, Anna Schapiro, and others for their comments on earlier drafts of this opinion article.
Publisher Copyright:
© 2019 Elsevier Ltd
PY - 2019/9
Y1 - 2019/9
N2 - What are the principles that govern whether neural representations move apart (differentiate) or together (integrate) as a function of learning? According to supervised learning models that are trained to predict outcomes in the world, integration should occur when two stimuli predict the same outcome. Numerous findings support this, but – paradoxically – some recent fMRI studies have found that pairing different stimuli with the same associate causes differentiation, not integration. To explain these and related findings, we argue that supervised learning needs to be supplemented with unsupervised learning that is driven by spreading activation in a U-shaped way, such that inactive memories are not modified, moderate activation of memories causes weakening (leading to differentiation), and higher activation causes strengthening (leading to integration).
AB - What are the principles that govern whether neural representations move apart (differentiate) or together (integrate) as a function of learning? According to supervised learning models that are trained to predict outcomes in the world, integration should occur when two stimuli predict the same outcome. Numerous findings support this, but – paradoxically – some recent fMRI studies have found that pairing different stimuli with the same associate causes differentiation, not integration. To explain these and related findings, we argue that supervised learning needs to be supplemented with unsupervised learning that is driven by spreading activation in a U-shaped way, such that inactive memories are not modified, moderate activation of memories causes weakening (leading to differentiation), and higher activation causes strengthening (leading to integration).
KW - differentiation
KW - fMRI
KW - integration
KW - neural networks
KW - unsupervised learning
UR - http://www.scopus.com/inward/record.url?scp=85069735584&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85069735584&partnerID=8YFLogxK
U2 - 10.1016/j.tics.2019.06.007
DO - 10.1016/j.tics.2019.06.007
M3 - Review article
C2 - 31358438
AN - SCOPUS:85069735584
SN - 1364-6613
VL - 23
SP - 726
EP - 742
JO - Trends in Cognitive Sciences
JF - Trends in Cognitive Sciences
IS - 9
ER -