Transcript Computational Explorations in Cognitive Neuroscience

Computational Cognitive
Neuroscience
Shyh-Kang Jeng
Department of Electrical Engineering/
Graduate Institute of Communication/
Graduate Institute of Networking and Multimedia
Artificial Intelligence
http://www.research.ibm.com/deepblue/meet/html/d.1.shtml
http://www.research.ibm.com/deepblue/press/html/g.6.6.shtml
羅仁權, 再造一個青年愛因斯坦, 台大科學創造新文明
特展, 2011
http://www.takanishi.mech.waseda.ac.jp/top/research/music/flute/wf_4rv/index_j.htm
Jeff Hawkins’s Comments on Artificial
Intelligence


AI defenders … a program that produces
outputs resembling (or surpassing) human
performance on a task in some narrow but
useful way really is just as good as the way
our brains do it
…this kind of ends-justify-the-means
interpretation of functionalism leads
AI researchers astray
J. Hawkins, On Intelligence, Times Books, 2004
Artificial Neural Networks
R. O. Duda, P. E. Harr, and D. G. Stork, Pattern Classification, 2nd ed., John Wiley & Sons, 2001
Jeff Hawkins’s Comments on
Artificial Neural Networks



Connectionists intuitively felt the brain wasn’t
a computer and that its secrets lie in how
neurons behave when connected together
That was a good start, but the field barely
moved on from its early successes
Research on cortically realistic networks was,
and remains, rare
Jeff Hawkins’s Comments on
Intelligence



Since intelligence is an internal property of a
brain, we have to look inside the brain to
understand what intelligence is
To succeed, we will need to crib heavily from
nature’s engine of intelligence, the neocortex
No other roads will get us there
Cognitive Neuroscience

To understand how neural processes give rise
to cognition


Perception, attention, language, memory, problem
solving, planning, reasoning, coordination and
execution of action
“Cognitive neuroscience – with its concern
about perception, action, memory, language,
and selective attention – will increasingly
come to represent the central focus of all
neurosciences in the twenty-first century.”
Experimental Methodologies

fMRI and other imaging modalities


Neural basis of cognition in human
Multi-electrode arrays


http://www.csulb.edu/~cwallis/482/fmri/fmri.h2.gif
Record from many separate neurons at a time
Insight into representation of information
http://paulbourke.net/oldstuff/eeg/eeg2.jpeg
http://en.wikibooks.org/wiki/File:Sleep_EEG_Stage_1.jpg
Other Major Research Methods

Processes occurring in individuals with
disorders



Helpful to understand the “normal” case
Animal models are also often used
Conscious experience



Subject to scientific scrutiny through observables
Including verbal reports or other readout methods
Brief interval of time or longer periods of time
Different Mechanistic Goals



Some focus on partitioning the brain into
distinct modules with isolable functions
Some try to find detailed characterization of
actual physical and chemical processes
Some look for something more general


Not the details themselves that matter
Principles that are embodied in these details are
more important
Two-Route Model for Reading
http://en.wikibooks.org/wiki/File:1_1_twoRouteModelInReading.JPG
Computational Cognitive Neuroscience


Understanding how the brain embodies the
mind, using biologically based computational
models made up of networks of neuron-like
units
Intersection of many disciplines



Neuroscience
Cognitive psychology
Computation
Computational Model for Reading
Randall C. O’Reilly and Yuko Munakata, Computational Explorations in Cognitive
Neuroscience: Understanding the Mind by Simulating the Brain, MIT Press, 2000
http://www.lps.uci.edu/~johnsonk/CLASSES/philpsych/brain.jpg
Randall C. O’Reilly and Yuko Munakata, Computational Explorations in Cognitive
Neuroscience: Understanding the Mind by Simulating the Brain, MIT Press, 2000
Usefulness of Models


Work through in detail of proposed modular
mechanism
Lead to


explicit predictions that can be compared for an
adequate account
exploration of what postulates imply about
resulting behaviors
Course Outline
Introduction and Overview
I. Basic Neural Computational Mechanisms
2. Individual Neurons
3. Networks of Neurons
4. Hebbian Model Learning
5. Error-Driven Task Learning
6. Combined Model and Task Learning
1.
Course Outline
II. Large-Scale Brain Area Organization and
Cognitive Phenomena
7. Large-Scale Brain Area Functional
Organization
8. Perception and Attention
9. Memory
10. Language
11. High-Level Cognition
Textbook and Website

Randall C. O’Reilly and Yuko Munakata,
Computational Explorations in Cognitive
Neuroscience: Understanding the Mind by
Simulating the Brain, MIT Press, 2000.

http://cc.ee.ntu.edu.tw/~skjeng/CCN2011.htm
Software Emergent
For practicing examples in the textbook and
doing homeworks as well as the term project
 Enhanced from PDP++
 Downloadable from
http://grey.colorado.edu/emergent/index.php/
Main_Page

http://grey.colorado.edu/emergent/index.php/File:Screenshot_ax_tutorial.png
References


Thomas J. Anastasio, Tutorial on Neural
Systems Modeling, Sinauer Associates Inc.
Publishers, 2010
Bernard J. Baars and Nicole M. Gage,
Cognition, Brain, and
Consciousness:Introduction to Cognitive
Neuroscience, 2nd ed., Academic Press, 2010
References


Friedemann Pulvermuller, The Neuroscience
of Language, Cambridge University Press,
2002
Douglas Medin, Brian H. Ross, Arthur B.
Markman, Cognitive Psychology, 4th ed,.
Wiley, 2004
References


Patricia Churchland and Terrence J.
Sejnowski, The Computational Brain
(Computational Neuroscience), MIT Press,
1994
Peter Dayan and L. F. Abbott, Theoretical
Neuroscience: Computational and
Mathematical Modeling of Neural Systems,
MIT Press, 2005
References

J. Hawkins, On Intelligence, Times Books,
2004