Robert, A. (1999); "Lamination and Within-Area Integration in
the Neocortex", Doctoral Dissertation, University of
California at San Diego, Cognitive Science Department.
My doctoral dissertation focuses on bridging from cortical anatomy to
computational models for Hebbian development in non-primary cortex. After a
review of neocortical evolution, inter-species and inter-area
cytoarchitectural variation, and major theories of the role feedforward and
feedback connections in sensory processing, it describes how I compiled cell
distribution and connection data focusing on rat primary sensory cortex and
built two detailed models from it using single-compartment spiking cells, both
of which were compared with slice experiments. The second model reduced the
number of layers and cell populations from the first, and I then simplified
this into an architecture with 3 (non-spiking) cell layers and a 3 x 3 matrix
of "lateral" interactions within and between them. Under Hebbian learning
algorithms similar to those employed for studying development of orientation
selectivity in primary visual cortex, I show that the 3-layer structure is
better able to integrate multiple inputs with different arbor widths and
strengths as are seen in real neocortical feedforward and feedback
connections. Dr. Martin
I. Sereno was my advisor.
(The above link is to a version formatted for two-sided printing. If you don't have access to an appropriate printer, please use this version (PDF) instead, which formats two-pages-on-one, to save a little paper.)
View abstract and introduction.
Robert, A. (1999); "Pyramidal Arborizations and Activity Spread in
Neocortex" in Neurocomputing 26-27:483-90.
This journal/conference paper provides a compressed version of the first third
or so of the work described above in connection with my dissertation. A
cortical model with 3 layers and 13 spiking cell populations was constructed
from anatomical data and compared with slice experiments. Interlayer
interaction functions suitable for use in non-spiking and Hebbian learning
models are calculated.
Robert, A. (2002); "Laminated Models of Within-Area Integration in the
Neocortex"; currently submitted.
This manuscript submitted for publication is a rather expanded version of the
Robert, A. (1998); "Blending and Other Conceptual Operations in the
Interpretation of Mathematical Proofs", in Koenig, J.P. (ed.),
Discourse and Cognition: Bridging the Gap, pp. 337-50.
This paper describes the application of notions from cognitive
semantics - image schemas and metaphorical mapping - to understanding
mathematical reasoning. Professor Gilles Fauconnier has played a
major role in helping me develop these ideas.
View extended abstract.
Robert, A. (2002); "Cognitive Structures and Processes in the Interp-
retation of Mathematical Proofs"; unpublished manuscript.
This is a much longer exposition of the work described in the paper above,
including several extensions. It was submitted to a journal but requires
revision. I hope to get to this in early 2003, but for now I am offering the
present version for anyone who is interested. Feedback, particularly from
practicing mathematicians, is welcome.
Robert, A. (1997); "From Contour Completion to Image Schemas: A Modern
Perspective on Gestalt Psychology", UCSD Department of Cognitive
Science Technical Report 9702, available from
This technical report assesses the Gestalt grouping laws and the principles
behind them in the light of experimental data and modeling results from modern
neuroscience. It suggests that although the grouping principles are in need
of revision, the general philosophy of the Gestalt approach -- that energy
minimization acting at the level of dynamic neural activity underlies a wide
range of psychological phenomena -- is valid and useful in the context of
contemporary cognitive science. In particular, the Gestalt theory is highly
compatible with present concepts and theoretical constructs in cognitive
semantics, itself a framework tying together diverse cognitive phenomena.
View abstract, outline, and conclusions.
See my research page for more information.
Back to Adrian's "serious" page.