Name: Randal A. Koene
Address: Neural Modeling Lab W7/13, Psychology [*]
1205 Dr.Penfield Avenue
Montreal, Quebec H3A 1B1, Canada
Phone: [*] (home)
(514) 398-4319 (lab)
E-mail: randalk*No*@*SpAm*marina.psych.mcgill.ca
(remove the "*No**SpAm*" notice)
Citizenship: Netherlands
Languages: English, German, Dutch, French
Education
2001 Ph.D, Graduation Date: June 30, 2001
Department of Psychology
McGill University
Thesis: "Multiphase Memory with Cycles of Spontaneous
Reactivation and Consolidation. The Increase of
Relational Complexity under Attentional Guidance."
GPA: 3.78
1995 Masters of Science in Electrical Engineering
Faculty of Electrical Engineering
Delft University of Technology
Thesis: "Extracting Knowledge in terms of Rules from
Trained Neural Networks"
GPA: 3.7
Awards and Honors
1996-1999 Max Stern Recruitment Fellowship
1995 Magna Cum Laude
Publications
1. Randal Koene and Yoshio Takane. Discriminant component pruning.
Regularization and interpretation of multi-layered back-propagation
networks. Neural Computation, Vol. 11, pp. 783-802, 1999.
2. Randal A. Koene. Improvements in Recall Behaviour of a Model Combining
Oscillatory Short Term Memory and Long Term Potentiation-Depression.
[Submitted to Learning and Memory, March 20, 2001]
3. Randal A. Koene. Localized Distributions of Connectivity and GABA_B
Control of Learning and Recall Modes in an Oscillatory Model of Memory.
[Submitted to Learning and Memory, March 20, 2001]
In Preparation
1. Randal A. Koene. Sparse Hippocampal Memory Encoding. Attentional
Gain-Control of Long-Term Memory Formation with Multiple Instantiation.
2. Randal A. Koene. Dense Neocortical Memory Consolidation. Selective Transfer
from Hippocampus with Multiple Instantiation through Recruitment.
3. Randal A. Koene. Multiphase Cortico-Hippocampal Memory. A Sequence of
Mechanisms Selective for Temporal Saliency leads to Long-Lasting
Consolidation and Integration of Memories.
Presentations
1. Randal Koene. The Extraction of Rules from Neural Networks with the
Knowledgetron Algorithm. University of Nijmegen Biophysics Meeting, 1996.
2. Randal Koene and Yoshio Takane. Discriminant Component Pruning: Effective
Regularization and Improved Interpretability of Back-propagation Neural
Networks. Brain, Behaviour and Cognitive Science Meeting, 1998.
3. Randal Koene. Attentional Highlighting and Neuron Recruitment in the
Transfer of Memories from Sparse Hippocampal to Dense Neocortical Encoding.
The Fifth International Conference on Cognitive and Neural Systems in
Boston, MA., June 2, 2001.
Pending Presentations
1. Randal Koene. Multiple instantiation in a model of short- and long-term
memory in the cortico-hippocampal loop. To be presented June 24, 2001 at
the Brain, Behaviour and Cognitive Science Conference in Laval, QC.
Research Experience
Neuroscience: synaptic modeling, dendritic operations, spiking neuron
models, compartmental neuronal modeling, PV-CG-CB interneuronal
function, large-scale parallel propagation of activity,
ion kinetics, LTP and LTD, oscillatory memory systems,
transfer of activity between oscillating networks,
synaptogenesis, neurogenesis and resource recruitment,
CREB and PKA synaptic modulation, spine growth, physiology of
long-term memory, brain rhythms and modalities, memory
consolidation and integration
Behavioural: signal analysis of multi-electrode recording, place cell
phenomena, path prediction and navigation
Cognitive: cognitive modeling in LISP, learning phenomena
Quantitative: information theoretic analysis, model reduction, parameter
significance evaluation, component discrimination and
identification, function approximation, splines, multi-variate
systems, differential analysis, biological modeling methodology
A.I.: distributed processes, knowledge engineering, fuzzy logic,
neural network control and pattern recognition, neuronal
modeling software library design, neuronal emulation algorithms
Optics and IC: optical modulation of signal transmission, experimental
microprocessor design
Teaching Experience
1. Delft University: Teaching Assistant
Software Engineering in Scheme/LISP, W1994, W1995
2. Delft University: Teaching Assistant
Electrical Engineering Lab I, S1994
3. McGill University: Teaching Assistant
Statistics for Experimental Design, F1996, W1997, F2000, W2001
Computer Skills
Languages: C, C++, Pascal, Fortran, Basic, Scheme, LISP, Assembler
various shell scripting languages
Operating Systems: Linux/Unix, Windows, Macintosh
Internet: HTML, CGI
Service
1. "Dispuut Infothedis" Student Association of the Information Theory Group,
student-staff relations, Chairman of Excursions, 1994, 1995
2. Graduate Association for Students in Psychology (GASP), VP Finance,
1996-2000
Current Research Project
My research has focused on the short-, intermediate- and long-term memory (STM,
ITM and LTM) processes of the cortico-hippocampal loop. First, I addressed the
postulate that attentional control kindles consolidation. This postulate may be
applicable as a selectivity function during several transitions from one stage
of human memory to another. A plausible model of STM and ITM has been proposed
by Jensen et al., but this model required some adaptation so that it could
operate reliably in a broader system. I successfully enhanced the model by
correcting the representation of the interneuronal network, as well as the LTP
and LTD synaptic update algorithm and the local distribution of connectivity. I
characterized the timing requirements for the synchronous cooperating of
multiple network stages exhibiting oscillations of coherent activity. My models
thereby achieved acquisition and ITM storage of episodes consisting of
arbitrarily large patterns of activation. I applied variable GABA_B modulation
to control learning and recall modes in models of dentate gyrus and
hippocampus. Learning and recall experiments with interacting patterns showed
that while STM could hold interacting patterns in consecutive gamma cycles,
cross-talk was a problem for synaptic storage. I discovered that association
with coincidentally coactivating cells undergoing pattern stimulus could
serve to multiplex originally interacting patterns into the sparse
representation hypothesized in the hippocampal system. This multiply
instantiated memory was flexible for rapid acquisition and therefore volatile.
I proceeded with the hypothesis that the attentional control manifested in
rapidly acquired ITM served to emphasize activity that can lead to robust LTM.
My model increased the likelihood of activity in target areas through
attentional highlighting in the form of LTP. By implementing synaptic
sensitivity to frequent activity and the presence of CREB, I achieved the
establishment of LTM traces at highlighted synapses.
In subsequent studies, I investigated the mechanistic requirements for the
establishment of long-lasting LTM and the integration of knowledge. Slow wave
sleep periods (Stickgold) and quiet waking periods (Hasselmo) have been
hypothesized to bring about a randomized and therefore interleaved reactivation
of memory patterns in the neocortex. I performed modeling studies that examined
the influence of such reactivation on consolidation. With this protocol, my
experiments accomplished the gradual transfer of memory representations from a
sparse hippocampal encoding to a dense neocortical encoding. I examined
selection criteria for the transfer and recruitment leading to long-lasting
LTM. I found that occasional reoccurrence of stimuli related to a memory
pattern caused potentiation at neocortical synapses that acted as attentional
gain-control for transfer from the hippocampus and consolidation in dense
encoding. In this dense LTM, my models achieved multiple instantiation to avoid
cross-talk between intersecting patterns by recruiting new cells into
coactivating clusters. Dense storage offered generalization capabilities and
the reuse of existing synaptic structures, while recruitment accomplished by
synaptogenesis onto existing cells or onto young cells generated by ongoing
neurogenesis (Gould) supported the retention of details in stored patterns.
Finally, I randomly reactivated patterns over prolonged periods during which
new knowledge was added to the networks. An analogous process is believed to
occur during REM as well as through coincidental external stimulation. This
resulted in novel associations that effectively broadened the distribution of
memory patterns. The experimental results resembled the progressive expansion
of memory distribution evidenced by correlations between the extent of
retrograde amnesia resulting from different degrees of damage to hippocampal
and surrounding areas. The discovery of new structure in data implied greater
integration. Summarizing, my research suggests that multiple memory stages with
selection criteria based on attentional gain-control mechanisms achieve storage
that is optimized for the duration of a memory's saliency and that maximizes
the integration of generally relevant knowledge.
References
1. Dr. Yoshio Takane
Professor of Quantitative Psychology
Department of Psychology, McGill University
1205 Dr. Penfield Avenue
Montreal, Quebec H3A 1B1, Canada
takane@takane2.psych.mcgill.ca
(514) 398-6125
2. Dr. Matthew Shapiro
Professor of Neurobiology
Neurobiology of Aging Laboratories
Research Center for Neurobiology
Mount Sinai School of Medicine
One Gustave L. Levy Place, Box 1639
New York, NY 10029-6574, U.S.A.
matthew.shapiro@mssm.edu
(212) 659-5941
3. Dr. Ronald Chase
Professor of Biology
Department of Biology, McGill University
1205 Dr. Penfield Avenue
Montreal, Quebec H3A 1B1, Canada
rchase@bio1.lan.mcgill.ca
(514) 398-6422
|
