?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Category+Induction+and+Representation&rft.creator=Harnad%2C+Stevan&rft.subject=Cognitive+Psychology&rft.subject=Neural+Nets&rft.subject=Philosophy+of+Language&rft.description=A+provisional+model+is+presented+in+which+categorical+perception%0A+++++++++++++++++++++++++++++++(CP)+provides+our+basic+or+elementary+categories.+In+acquiring+a+category+we%0A+++++++++++++++++++++++++++++++learn+to+label+or+identify+positive+and+negative+instances+from+a+sample+of%0A+++++++++++++++++++++++++++++++confusable+alternatives.+Two+kinds+of+internal+representation+are+built+up+in+this%0A+++++++++++++++++++++++++++++++learning+by+%22acquaintance%22%3A+(1)+an+iconic+representation+that+subserves+our%0A+++++++++++++++++++++++++++++++similarity+judgments+and+(2)+an+analog%2Fdigital+feature-filter+that+picks+out+the%0A+++++++++++++++++++++++++++++++invariant+information+allowing+us+to+categorize+the+instances+correctly.+This%0A+++++++++++++++++++++++++++++++second%2C+categorical+representation+is+associated+with+the+category+name.%0A+++++++++++++++++++++++++++++++Category+names+then+serve+as+the+atomic+symbols+for+a+third+representational%0A+++++++++++++++++++++++++++++++system%2C+the+(3)+symbolic+representations+that+underlie+language+and+that%0A+++++++++++++++++++++++++++++++make+it+possible+for+us+to+learn+by+%22description.%22+Connectionism+is+one%0A+++++++++++++++++++++++++++++++possible+mechainsm+for+learning+the+sensory+invariants+underlying%0A+++++++++++++++++++++++++++++++categorization+and+naming.+Among+the+implications+of+the+model+are+(a)+the%0A+++++++++++++++++++++++++++++++%22cognitive+identity+of+(current)+indiscriminables%22%3A+Categories+and+their%0A+++++++++++++++++++++++++++++++representations+can+only+be+provisional+and+approximate%2C+relative+to+the%0A+++++++++++++++++++++++++++++++alternatives+encountered+to+date%2C+rather+than+%22exact.%22+There+is+also+(b)+no%0A+++++++++++++++++++++++++++++++such+thing+as+an+absolute+%22feature%2C%22+only+those+features+that+are+invariant%0A+++++++++++++++++++++++++++++++within+a+particular+context+of+confusable+alternatives.+Contrary+to+prevailing%0A+++++++++++++++++++++++++++++++%22prototype%22+views%2C+however%2C+(c)+such+provisionally+invariant+features+must%0A+++++++++++++++++++++++++++++++underlie+successful+categorization%2C+and+must+be+%22sufficient%22+(at+least+in+the%0A+++++++++++++++++++++++++++++++%22satisficing%22+sense)+to+subserve+reliable+performance+with+all-or-none%2C%0A+++++++++++++++++++++++++++++++bounded+categories%2C+as+in+CP.+Finally%2C+the+model+brings+out+some+basic%0A+++++++++++++++++++++++++++++++limitations+of+the+%22symbol-manipulative%22+approach+to+modeling+cognition%2C%0A+++++++++++++++++++++++++++++++showing+how+(d)+symbol+meanings+must+be+functionally+grounded+in%0A+++++++++++++++++++++++++++++++nonsymbolic%2C+%22shape-preserving%22+representations+--+iconic+and+categorical%0A+++++++++++++++++++++++++++++++ones.+Otherwise%2C+all+symbol+interpretations+are+ungrounded+and%0A+++++++++++++++++++++++++++++++indeterminate.+This+amounts+to+a+principled+call+for+a+psychophysical+(rather%0A+++++++++++++++++++++++++++++++than+a+neural)+%22bottom-up%22+approach+to+cognition.&rft.publisher=Cambridge+University+Press&rft.contributor=Harnad%2C+Stevan&rft.date=1987&rft.type=Book+Chapter&rft.type=NonPeerReviewed&rft.format=text%2Fhtml&rft.identifier=http%3A%2F%2Fcogprints.org%2F1572%2F1%2Fharnad87.categorization.html&rft.identifier=++Harnad%2C+Stevan++(1987)+Category+Induction+and+Representation.++%5BBook+Chapter%5D+++++&rft.relation=http%3A%2F%2Fcogprints.org%2F1572%2F