Learned Categorical Perception in Neural Nets: Implications for Symbol Grounding
| Abstract | After people learn to sort objects into categories they see them differently. Members of the same category look more alike and members of different categories look more different. This phenomenon of within-category compression and between-category separation in similarity space is called categorical perception (CP). It is exhibited by human subjects, animals and neural net models. In backpropagation nets trained first to auto-associate 12 stimuli varying along a onedimensional continuum and then to sort them into 3 categories, CP arises as a natural side-effect because of four factors: (1) Maximal interstimulus separation in hidden-unit space during autoassociation learning, (2) movement toward linear separability during categorization learning, (3) inverse-distance repulsive force exerted by the between-category boundary, and (4) the modulating effects of input iconicity, especially in interpolating CP to untrained regions of the continuum. Once similarity space has been "warped" in this way, the compressed and separated "chunks" have symbolic labels which could then be combined into symbol strings that constitute propositions about objects. The meanings of such symbolic representations would be "grounded" in the system's capacity to pick out from their sensory projections the object categories that the propositions were about. | |||||||||
| Keywords | No keywords specified (fix it) | |||||||||
| Categories | No categories specified (fix it) | |||||||||
| Options |
|
|||||||||
Download options| PhilPapers Archive |
Upload a copy of this paper Check publisher's policy on self-archival Papers currently archived: 5,709 |
| External links |
  Try with proxy. |
| Through your library | Only published papers are available at libraries |
Similar books and articlesStevan Harnad (1992). Connecting Object to Symbol in Modeling Cognition. In A. Clark & Ronald Lutz (eds.), Connectionism in Context. Springer-Verlag.
Stevan Harnad, Symbol Grounding is an Empirical Problem: Neural Nets Are Just a Candidate Component.
Stevan Harnad (1995). Does Mind Piggyback on Robotic and Symbolic Capacity? In H. Morowitz & J. Singer (eds.), The Mind, the Brain, and Complex Adaptive Systems. Addison Wesley.
Stevan Harnad (1987). Category Induction and Representation. In [Book Chapter].
Stevan Harnad (2002). Symbol Grounding and the Origin of Language. In Matthias Scheutz (ed.), Computationalism: New Directions. MIT Press.
Yasmina Jraissati (forthcoming). Categorical Perception of Color: Assessing the Role of Language. Croatian Journal of Philosophy.
Constantine Tsinakis & Han Zhang (2004). Order Algebras as Models of Linear Logic. Studia Logica 76 (2):201 - 225.
Yasmina Jraissati (2012). Categorical Perception of Color. Croatian Journal of Philosophy 12 (3):439-462.
John E. Hummel (2010). Symbolic Versus Associative Learning. Cognitive Science 34 (6):958-965.
Analytics
Monthly downloads |
Added to index2010-12-22Total downloads4 ( #178,844 of 550,860 )Recent downloads (6 months)1 ( #63,425 of 550,860 )How can I increase my downloads? |
My notes
Discussion
