U.S. patent application number 12/421859 was filed with the patent office on 2011-07-28 for mobile brain-based device for use in a real world environment.
This patent application is currently assigned to Neurosciences Research Foundation, Inc.. Invention is credited to Gerald M. Edelman, Jeffrey L. Krichmar, Jeffrey L. McKinstry, Anil K. Seth.
Application Number | 20110184556 12/421859 |
Document ID | / |
Family ID | 35242324 |
Filed Date | 2011-07-28 |
United States Patent
Application |
20110184556 |
Kind Code |
A1 |
Seth; Anil K. ; et
al. |
July 28, 2011 |
MOBILE BRAIN-BASED DEVICE FOR USE IN A REAL WORLD ENVIRONMENT
Abstract
A mobile brain-based device BBD includes a mobile base equipped
with sensors and effectors (Neurally Organized Mobile Adaptive
Device or NOMAD), which is guided by a simulated nervous system
that is an analogue of cortical and sub-cortical areas of the brain
required for visual processing, decision-making, reward, and motor
responses. These simulated cortical and sub-cortical areas are
reentrantly connected and each area contains neuronal units
representing both the mean activity level and the relative timing
of the activity of groups of neurons. The brain-based device BBD
learns to discriminate among multiple objects with shared visual
features, and associated "target" objects with innately preferred
auditory cues. Globally distributed neuronal circuits that
correspond to distinct objects in the visual field of NOMAD 10 are
activated. These circuits, which are constrained by a reentrant
neuroanatomy and modulated by behavior and synaptic plasticity,
result in successful discrimination of objects. The brain-based
device BBD is moveable, in a rich real-world environment involving
continual changes in the size and location of visual stimuli due to
self-generated or autonomous, movement, and shows that reentrant
connectivity and dynamic synchronization provide an effective
mechanism for binding the features of visual objects so as to
reorganize object features such as color, shape and motion while
distinguishing distinct objects in the environment.
Inventors: |
Seth; Anil K.; (San Diego,
CA) ; McKinstry; Jeffrey L.; (San Diego, CA) ;
Edelman; Gerald M.; (La Jolla, CA) ; Krichmar;
Jeffrey L.; (Cardiff-by-the-Sea, CA) |
Assignee: |
Neurosciences Research Foundation,
Inc.
San Diego
CA
|
Family ID: |
35242324 |
Appl. No.: |
12/421859 |
Filed: |
April 10, 2009 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11105019 |
Apr 13, 2005 |
7519452 |
|
|
12421859 |
|
|
|
|
60562376 |
Apr 15, 2004 |
|
|
|
Current U.S.
Class: |
700/249 |
Current CPC
Class: |
G06N 3/004 20130101;
G16H 50/50 20180101; G06N 3/063 20130101; G06F 3/015 20130101; G06N
3/049 20130101; G06N 20/00 20190101; G06K 9/629 20130101 |
Class at
Publication: |
700/249 |
International
Class: |
G06F 15/18 20060101
G06F015/18 |
Goverment Interests
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH AND
DEVELOPMENT
[0002] This invention was made with Government support under
N00014-03-1-0980 awarded by the Office of Naval Research. The
United States Government has certain rights in the invention.
Claims
1-13. (canceled)
14. A mobile brain-based device (BBD) for behaving in a real-world
environment to integrate a visual scene, comprising: a. a mobile
adaptive device having i. a visual input sensor for receiving
visual information; and ii. an effector for enabling movement of
said mobile adaptive device; b. a computer-based simulated nervous
system modeling the regional and functional neuro-anatomy of the
cortical regions of a human brain for visually recognizing and
discriminating between different objects within the visual scene,
said computer-based simulated nervous system including i. a first
neural area forming a visual system and responsive to visual input
from said visual input sensor for producing visual stimuli, said
first neural area corresponding to the ventral cortical pathway of
the brain for producing visual stimuli; ii. a second neural area,
analogous to an ascending neuromodulatory system, responsive to a
real-world salient event experienced by the mobile brain-based
device while being mobile in its real-world environment, for
producing value stimuli; and iii. a third neural area,
corresponding to the superior colliculus area of the brain and
responsive to said visual and value stimuli, for controlling said
effector to orient said mobile adaptive device towards the visual
input information to said mobile adaptive device; c. wherein visual
recognition and discriminating between different objects is
achievable during real-world mobility of said mobile adaptive
device through reentrant connectivity of neuronal units within each
of said first, second and third neural areas, through reentrant
connectivity between said first, second and third neural areas, and
through the interaction of local processes, which are activities
within each of said first, second and third neural areas, and
global processes which create functional neural circuits formed
during the real-world operation and having synchronous activity
between said first, second and third neural areas; d. wherein
connectivity between said first neural area and said second neural
area are value-dependent synaptic plastic connections; e. wherein
connectivity from said first neural area to said third neural area
are value-dependent synaptic plastic connections; and f. wherein
connectivity from said second neural area to said third neural area
are excitatory synaptic plastic connections.
15. A mobile brain-based device according to claim 14, wherein each
said first, second and third neural areas has neuronal units, and
wherein said neuronal units in each said area have relative
neuronal activity whose timing is represented by a firing rate
variable and the relative timing of which is represented by a phase
variable, in which similar firing phases of neuronal units in said
areas reflect synchronous activity.
16. A mobile brain-based device according to claim 14, wherein said
value stimuli modify the strength of the synaptic plastic
connections between said first, second and third neural areas to
provide for the adaptive behavior of the mobile brain-based device
in a real-world environment.
17. A mobile brain-based device according to claim 14, wherein said
first neural area corresponds to said vertical cortical pathway
having neural areas V1, V2, V4 and IT being coupled in a pathway
V1.fwdarw.V2.fwdarw.V4.fwdarw.IT.
18. A mobile brain-based device according to claim 14, wherein each
of said first, second and third neural areas includes neuronal
units, in which said neuronal units have excitatory synaptic
connections amongst themselves, and each of said excitatory
synaptic connections are voltage-dependent.
19. A mobile brain-based device according to claim 18, wherein said
first, second and third neural areas have reentrant excitatory
connections between said areas, and all said reentrant excitatory
connections are voltage-dependent.
Description
PRIORITY CLAIM
[0001] This application is a continuation of U.S. patent
application Ser. No. 11/105,019, filed Apr. 13, 2005, entitled
"Mobile Brain-Based Device for Use in a Real World Environment," by
Anil K. Seth et al., which claims priority under 35 U.S.C. 119(e)
to U.S. Provisional Patent Application No. 60/562,376, filed Apr.
15, 2004, entitled "Mobile Brain-Based Device for Use in a Real
World Environment," by Anil K. Seth et al., which applications are
incorporated herein by reference.
COPYRIGHT NOTICE
[0003] A portion of the disclosure of this patent document contains
material which is subject to copyright protection. The copyright
owner has no objection to the facsimile reproduction by anyone of
the patent document or the patent disclosure, as it appears in the
Patent and Trademark Office patent file or records, but otherwise
reserves all copyright rights whatsoever.
FIELD OF THE INVENTION
[0004] The present invention relates to the field of brain-based
devices having simulated nervous systems for guiding the behavior
of the devices in a real world environment.
BACKGROUND OF THE INVENTION
[0005] A brain-based device is a device that has a sensing system
for receiving information, effectors that enable the device to move
about, and a simulated nervous system which controls movement of
the effectors in response to input from the sensing system to guide
the behavior of the brain-based device in a real-world environment.
The sensing system may include video and audio sensors which
receive image and audio information from the real-world environment
in which the device moves. The simulated nervous system may be
implemented as a computer-based system which receives and processes
the image and auditory information input to the brain-based device
and outputs commands to the effectors to control the behavior of
the device in the environment.
[0006] The simulated nervous system, while implemented in a
computer-based system, emulates the human brain rather than a
programmed computer which typically follows a set of precise
executable instructions or which performs computations. That is,
the brain is not a computer and follows neurobiological rather than
computational principles in its construction. The brain has special
features or organization and functions that are not believed to be
consistent with the idea that it follows such a set of precise
instructions or that it computes in the manner of a programmed
computer. A comparison of the signals that a brain receives with
those of a computer shows a number of features that are special to
the brain. For example, the real world is not presented to the
brain like a data storage medium storing an unambiguous series of
signals that are presented to a programmed computer. Nonetheless,
the brain enables humans (and animals) to sense their environment,
categorize patterns out of a multitude of variable signals, and
initiate movement. The ability of the nervous system to carry out
perceptual categorization of different signals for sight, sound,
etc. and divide them into coherent classes without a prearranged
code is special and unmatched by present day computers, whether
based on artificial intelligence (AI) principles or neural network
constructions.
[0007] The visual system of the brain contains a variety of
cortical regions which are specialized to different visual
features. For example, one region responds to the color of an
object, another separate region responds to the object's shape,
while yet another region responds to any motion of the object. The
brain will enable a human to see and distinguish in a scene, for
example, a red airplane from a gray cloud both moving across a
background of blue sky. Yet, no single region of the brain has
superordinate control over the separate regions responding to
color, shape and movement that coordinate color, shape and movement
so that we see and distinguish a single object (e.g. the airplane)
and distinguish it from other objects in the scene (e.g. the cloud
and the sky).
[0008] The fact that there is no such single superordinate control
region in the brain poses what is known as the "binding problem."
How do these functionally separated regions of the brain coordinate
their activities in order to associate features belonging to
individual objects and distinguish among different objects? It is
this ability of the brain to so associate and distinguish different
objects that enables us to move about in our real-world
environment. A mobile brain-based device having a simulated nervous
system that can control the behavior of the device in the rich
environment of the real world therefore would have many advantages
and uses.
[0009] Mechanisms proposed for solving the "binding problem"
generally fall into one of two classes: (i) binding through the
influence of "higher" attentional mechanisms of the brain, and (ii)
selective synchronization of the "firing" of dynamically formed
groups of neurons in the brain. In (i), the belief is that the
brain through its parietal or frontal regions, "binds" objects by
means of an executive mechanism, for example, a spotlight of
attention that would combine visual features appearing at a single
location in space, e.g. the red airplane or gray cloud against the
background of a blue sky. In (ii), the belief is that the brain
"binds" objects in an automatic, dynamic, and pre-attentive process
through groups of neurons that become linked by selective
synchronization of the firing of the neurons. These synchronized
neuronal groups form within the brain into global patterns of
activity, or circuits, corresponding to perceptual categories. This
enables us to see, for example, a red, flying airplane as a single
object distinct from other objects such as a gray, moving
cloud.
[0010] Computer-based computational models of visual binding, as
well as physical, mobile brain-based devices having a simulated
nervous system, are known, Yet, neither provides emergent circuits
in the computer model or in the simulated nervous system of the
physical brain-based device that contribute to providing a device
with a rich and variable behavior in the real-world environment,
especially in environments that require preferential behavior
towards one object among many in a scene. For example, it would be
desirable to have a mobile brain-based device move about in an
environment and have preferential behavior toward one object among
many in a scene so as to be able to obtain images of that object
via an on-board camera and to select that object via on-board
grippers.
[0011] One prior computational computer model simulated the nervous
system by representing nine neural areas analogous to nine cortical
areas of the visual system of the brain. It also simulated "reward"
and motor systems of the nervous system. The model had "reentrant
connections" or circuits between the nine different cortical areas,
which are connections that allow the cortical areas to interact
with each other. This computational model showed the capabilities
of reentrant circuits to result in binding; the computer model,
however, had several limitations. The stimuli into the modeled
nervous system came from a limited predefined set of simulated
object shapes and these were of uniform scale, contrary to what is
found in a real-world environment. Furthermore, the resulting
modeled behavior did not emerge in a rich and noisy environment
experienced by behaving organisms in the real world. A more
detailed description of this computational model is given in the
paper entitled "Reentry and the Problem of Integrating Multiple
Cortical Areas: Simulation of Dynamic Integration in the Visual
System", by Tononi and Edelman, Cerebral Cortex, July/August
1992.
[0012] A prior physical, mobile brain-based device having a
simulated nervous system does explore its environment and through
this experience learns to develop adaptive behaviors. Such a prior
mobile brain-based device is guided by the simulated nervous system
which is implemented on a computer system. The simulation of the
nervous system was based on the anatomy and physiology of
vertebrate nervous systems, but as with any simulated nervous
system, with many fewer neurons and a simpler architecture than is
found in the brain. For this physical, mobile brain-based device,
the nervous system was made up of six major neural areas analogous
to the cortical and subcortical brain regions. These six major
areas included: an auditory system, a visual system, a taste
system, a motor system capable of triggering behavior, a visual
tracking system, and a value system. A detailed description of this
mobile brain-based device is given in the paper entitled "Machine
Psychology: Autonomous Behavior, Perceptual Categorization and
Conditioning in a Brain-based Device" by Krichmar and Edelman,
Cerebral Cortex, August 2002. While this brain-based device does
operate in a real-world environment, it does not implement, among
many other things, reentrant connections, thereby limiting its
ability to engage in visually guided behavior and in object
discrimination in a real-world environment.
SUMMARY OF THE INVENTION
[0013] The present invention is a physical, mobile brain-based
device ("BBD") having a simulated nervous system for guiding the
device in a rich exploratory and selective behavior in a real-world
environment. The simulated nervous system of this device contains
simulated neural areas analogous to the ventral stream of a brain's
visual system, known as neural areas V1 V2, V4 and IT that
influence visual tracking (neural area C), and neural areas having
a value system (area S). These neural areas have reentrant
connections within and between each other, which give rise to
biases in motor activity, which in turn evoke behavioral responses
in the mobile device enabling visual object discrimination in a
scene.
[0014] Each neural area is comprised of many neuronal units. And,
to represent the relative timing of neuronal activity, each
neuronal unit in each neural area is described by a firing rate
variable and a phase variable, where similar phases reflect
synchronous firing. The binding problem, therefore, in the present
invention is resolved based on principles of reentrant connectivity
and synchronous neuronal firing.
[0015] The physical, mobile device of the present invention, as it
is moving and interacting in the real world in a conditioning or
training stage, learns what objects are in its environment, i.e.
objects are not given to it as predefined data in a simulation.
That is, the brain-based device of the present invention learns, in
a given environment, what is a particular object, such as a green
diamond, what is a floor, what is a wall, etc. Moreover, this
learning through movement and interaction in the environment
results in the brain-based device having invariant object
recognition. This means that once it learns what, for example, a
green diamond is as an object during a training stage, it will
recognize that object when in a testing stage as the device moves
about its real-world environment whether the object is across a
room from the device, directly in front of the device, off to the
left of the device, off to the right of the device, etc.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] FIG. 1 is a pictorial view of a physical, mobile brain-based
device.
[0017] FIG. 2 is a schematic of the regional and functional
neuroanatomy of the simulated nervous system of the brain-based
device of FIG. 1.
[0018] FIGS. 3A-3E are views of a simple network of three neuronal
units used to explain the neuroanatomy of the present invention
shown in FIG. 2.
[0019] FIGS. 4A-4E illustrate graphically activity vs. phase of a
neuronal unit.
[0020] FIGS. 5A-5B illustrates schematically and photographically,
respectively, an experimental set-up of a real-world environment in
which the mobile brain-based device of FIGS. 1 and 2 behaves.
[0021] FIGS. 6A-6B are used to explain, respectively, the training
and testing protocol of the brain-based device of FIGS. 1 and
2.
[0022] FIGS. 7A-7B are graphs illustrating the behavior of the
brain-based device of FIGS. 1 and 2 following conditioning.
[0023] FIG. 8 is a snapshot of the neuronal unit activity of the
brain-based device of FIGS. 1 and 2 during a behavioral
experiment.
[0024] FIGS. 9A-9B show phase responses with and without reentry
connections, respectively, of the brain-based device of FIGS. 1 and
2 following conditioning.
[0025] FIGS. 10A-10C are used to explain phase correlations among
neural areas for the brain-based device of FIGS. 1 and 2
conditioned to prefer or discriminate a red diamond as a single
target object.
[0026] FIGS. 11A-11B are illustrations used to explain the response
of the neural value area S to target objects in different
real-world positions and at different scales.
[0027] FIGS. 12A-12B show neural activity during conditioning or
training of the brain-based device of FIGS. 1 and 2 for neural
areas S, IT and C.
[0028] FIG. 13 is an exemplary illustration of a system in
accordance with various embodiments of the invention.
[0029] FIG. 14 is a flow diagram illustration of neural simulator
initialization.
[0030] FIG. 15 is a flow diagram illustration of the master
component in accordance with various embodiments of the
invention.
[0031] FIG. 16 is a flow diagram illustration of a neural simulator
in accordance with various embodiments of the invention.
DETAILED DESCRIPTION
[0032] Aspects of the invention are illustrated by way of example
and not by way of limitation in the figures of the accompanying
drawings in which like references indicate similar elements. It
should be noted that references to "an", "one" and "various"
embodiments in this disclosure are not necessarily to the same
embodiment, and such references mean at least one. In the following
description, numerous specific details are set forth to provide a
thorough description of the invention. However, it will be apparent
to one skilled in the art that the invention may be practiced
without these specific details. In other instances, well-known
features have not been described in detail so as not to obscure the
invention.
[0033] FIG. 1 is a pictorial view of a brain-based device (BBD) of
the present invention which includes physically instantiated mobile
Neurally Organized Mobile Adaptive Device (NOMAD) 10 which can
explore its environment and develop adaptive behavior while
experiencing it. The brain-based device BBD also includes a
simulated nervous system 12 (FIG. 2) for guiding NOMAD 10 in its
real-world environment. In one embodiment, the simulated nervous
system 12, as will be further described, can run on a cluster of
computer workstations (see FIG. 13) remote from NOMAD 10. In this
embodiment, NOMAD 10 and the computer workstations communicate with
one another via wireless communication, thereby enabling untethered
exploration of NOMAD 10.
[0034] NOMAD 10 develops or adapts its behavior by learning about
the environment using the simulated nervous system 12. As NOMAD 10
moves autonomously in its environment, it will approach and view
multiple objects that share visual features, e.g. same color, and
have distinct visual features such as shape, e.g. red square vs.
red triangle. NOMAD 10 can become conditioned through the learning
experience to prefer one target object, e.g. the red diamond, over
multiple distracters or non-target objects such as the red square
and a green diamond of a scene in its vision. NOMAD 10 learns this
preference behaviorally while moving in its environment by
orienting itself towards the target object in response to an
audible tone.
[0035] NOMAD 10 has a CCD camera 16 for vision and microphones 18,
20 on either side of camera 16, which can provide visual and
auditory sensory input to simulated nervous system 12, as well as
effectors or wheels 22 for movement. It also has an infrared (IR)
sensor 24 at the front of NOMAD 10 for obstacle avoidance by
sensing differences in reflectivity of the surface on which it
moves, and for triggering reflexive turns of NOMAD 10 in its
environment. NOMAD 10 also contains a radio modem to transmit
status, IR sensor information, and auditory information to the
computer workstation carrying out the neural simulation via
simulated nervous system 12 and to receive motor commands from the
simulated nervous system 12 to control effectors 22. Video output
from camera 16 can be sent to the computer workstations via RF
transmission. All behavioral activity of NOMAD 10, other than the
IR reflexive turns, is evoked by signals received from simulated
nervous system 12.
[0036] FIG. 2 is a schematic diagram of the regional and functional
neuroanatomy of simulated nervous system 12 which guides the
behavior of NOMAD 10 in its environment. Simulated nervous system
12 is modeled on the anatomy and physiology of the mammalian
nervous system but, as can be appreciated, with far fewer neurons
and a much less complex architecture. Simulated nervous system 12
includes a number of neural areas labeled according to the
analogous cortical and subcortical regions of the human brain.
Thus, FIG. 2 shows respective neural areas labeled as V1, V2, V4,
IT, S, A-left, Mic-left, A-right, Mic-right and C, whose activity
controls the tracking of NOMAD 10. Each neural area V1, V2, etc.
contains different types of neuronal units, each of which
represents a local population of neurons. Each ellipse shown in
FIG. 2 (except "Tracking") denotes a different neural area, with
each such area having many neuronal units. To distinguish modeled
or simulated neural areas from corresponding regions in the
mammalian nervous system, the simulated areas are indicated in
italics, e.g. IT.
[0037] The neuroanatomy of FIG. 2 also shows schematically various
projections P throughout the simulated nervous system 12. A
projection can be "feedforward" from one neural area to another,
such as the projection P1 from neural area V1 to neural area V2. A
projection P may also be "reentrant" between neural areas such as
the reentrant projection P2 from neural area IT to neural area V4
and reentrant projection P4 from neural area V4 to neural area V2.
Reentrant projections P marked with an "X" were removed from the
simulated nervous system 12 during "lesion" experiments as will be
further described. Furthermore, projections P have properties as
indicated by the legend in FIG. 2, which are (1) "excitatory
voltage independent", (2) "excitatory voltage dependent", (3)
"plastic", (4) "inhibitory," and (5) "value dependent."
[0038] The simulated nervous system 12 shown in FIG. 2 is comprised
of four systems: a visual system, a tracking system, an auditory
system and a value system.
FIG. 2--Visual System. Neural Areas V1, V2, V4, IT
[0039] The visual system is modeled on the primate occipitotemporal
or ventral cortical pathway and includes neural areas
V1.fwdarw.V2.fwdarw.V4.fwdarw.IT in which neurons in successive
areas have progressively larger receptive fields until, in
inferotemporal cortex, receptive fields cover nearly the entire
visual field. Visual images from the CCD camera 16 of NOMAD 10 are
filtered for color and edges and the filtered output directly
influences neural activity in area V1. V1 is divided into
subregions (not shown) each having neuronal units that respond
preferentially to green (V1-green), red (V1-red), horizontal line
segments (V1-horizontal), vertical line segments (V1-vertical),
45-degree lines (V1-diagonal-right), and 135-degree lines
(V1-diagonal-left). This visual system provides a computationally
tractable foundation for analyzing higher-level interactions within
the visual system and between the visual system and other cortical
areas.
[0040] Subregions of neural area V1 project topographically to
corresponding subregions of neural area V2. The receptive fields of
neuronal units in area V2 are narrow and correspond closely to
pixels from the image of CCD camera 16. Neural area V2 has both
excitatory and inhibitory reentrant connections within and among
its subregions. Each V2 subregion projects to a corresponding V4
subregion topographically but broadly, so that neural area V4's
receptive fields are larger than those of neural area V2. Neural
area V4 subregions project back to the corresponding neural area V2
subregions with non-topographic reentrant connections. The
reentrant connectivity within and among subregions of area V4 is
similar to that in area V2. V4 projects in turn non-topographically
to neural area IT so that each neuronal unit in neural area IT
receives input from three V4 neuronal units randomly chosen from
three different V4 subregions. Thus, while neuronal units in IT
respond to a combination of visual inputs, the level of synaptic
input into a given IT neuronal unit is fairly uniform; this
prevents the activity of individual IT neuronal units from
dominating the overall activity patterns. IT neuronal units project
to other IT neuronal units through plastic connections, and back to
neural area V4 through non-topographic reentrant connections.
FIG. 2--Tracking System--Neural Area C
[0041] The tracking system allows NOMAD 10 to orient towards
auditory and visual stimuli. The activity of neural area C
(analogous to the superior colliculus) dictates where NOMAD 10
directs its camera gaze. Tracking in NOMAD 10 is achieved by
signals to wheels 22 based on the vector summation of the activity
of the neuronal units in area C. Each neuronal unit in area C has a
receptive field which matches its preferred direction, and the area
has a topographic arrangement such that if activity is
predominately on the left side of area C, signals to NOMAD 10's
wheels 22 are issued that evoke a turn towards the left. The
auditory neural areas (A-left and A-right) have strong excitatory
projections to the respective ipsilateral sides of area C causing
NOMAD 10 to orient towards a sound source. Neural area V4 projects
topographically to area C, its activity causing NOMAD 10 to center
its gaze on a visual object (e.g. a red triangle). Both neural
areas IT and the value system S project to area C, and plastic
connections in the pathways IT.fwdarw.C and IT.fwdarw.S facilitate
target selection by creating a bias in activity, reflecting salient
perceptual categories (see Value System, below). As will be
described below, prior to a conditioning or training stage, because
of a lack of bias, NOMAD 10 will direct its gaze predominately
between two objects in its environment (e.g. a red triangle and a
red square). After learning to prefer a visual object (e.g. a red
triangle), changes in the strengths of the plastic connections
result in greater activity in those parts of area C corresponding
to the preferred object's position.
FIG. 2--Auditory System--Neural Areas Mic-Left, Mic-Right, A-Left,
A-Right
[0042] This system converts inputs from microphones 16,18 into
simulated neuronal unit activity. Neural areas Mic-left and
Mic-right are respectively activated whenever the corresponding
microphones 16, 18 detect a sound of sufficient amplitude within a
specified frequency range. Mic-left/Mic-right project to neuronal
units in areas A-left/A-right. Sound from one side results in
activity on the ipsilateral side of the auditory system, which in
turn produces activity on the ipsilateral side of area C causing
orientation of NOMAD 10 towards the sound source.
FIG. 2--Value System--Neural Area S
[0043] Activity in the simulated value system signals the
occurrence of salient sensory events and this activity contributes
to the modulation of connection strengths in pathways IT.fwdarw.S
and IT.fwdarw.C. Initially, in the learning stage to be described
below, neural area S is activated by sounds detected by auditory
system (see A-left.fwdarw.S and A-right.fwdarw.S of nervous system
12). Activity in area S is analogous to that of ascending
neuromodulatory systems in that it is triggered by salient events,
influences large regions of the simulated nervous system (described
below in the section Synaptic Plasticity), and persists for several
cycles. In addition, due to its projection to the tracking area C,
area S has a direct influence on the behavior of NOMAD 10 in its
real-world environment.
[0044] Details of the values of certain parameters of the neuronal
units within the respective neural areas V1, V2, etc. shown in FIG.
2 are given in Table 1, described below. Details of the anatomical
projections and connection types of neuronal units of the neural
areas V1, V2, etc. are given in Table 2, described below. As is
known, a neuronal unit can be considered pre- or post- a synapse
(see "A Universe of Consciousness", by Edelman and Tononi, Basic
Books, 2000, FIG. 4.3, for a description of a synapse and pre- and
post-synaptic neurons.) The simulated nervous system 12 used in the
experiments described below contains 28 neural areas V1, V2, etc.,
53,450 neuronal units, and approximately 1.7 million synaptic
connections.
Neuronal Units--Generally
[0045] In one embodiment, a neuronal unit within a neural area V1,
V2, etc. of the simulated nervous system 12 is simulated by a mean
firing rate model. The state of each neuronal unit is determined by
both a mean firing rate variable (.sigma.) and a phase variable
(P). The mean firing rate variable of each neuronal unit
corresponds to the average activity or firing rate of a group of
roughly 100 neurons during a time period of approximately 100
milliseconds. The phase variable, which specifies the relative
timing of firing activity, provides temporal specificity without
incurring the computational costs associated with modeling of the
spiking activity of individual neurons in real-time (see Neuronal
Unit Activity and Phase, below).
Synaptic Connections--Generally
[0046] In one embodiment, synaptic connections between neuronal
units, both within a given neural area, e.g. V1 or C, and between
neural areas, e.g. V2.fwdarw.V4 or C.fwdarw.V4, are set to be
either voltage-independent or voltage-dependent, either
phase-independent or phase-dependent, and either plastic or
non-plastic, as indicated by the legend in FIG. 2.
Voltage-independent connections provide synaptic input to a
post-synaptic neuron regardless of the post-synaptic state of the
neuron. Voltage-dependent connections represent the contribution of
receptor types (e.g. NMDA receptors) that require post-synaptic
depolarization to be activated. In other words, a pre-synaptic
neuron will send a signal along its axon through a synapse to a
post-synaptic neuron. The post-synaptic neuron receives this signal
and integrates it with other signals being received from other
pre-synaptic neurons.
[0047] A voltage independent connection is such that if a
pre-synaptic neuron is firing at a high rate, then a post-synaptic
neuron connected to it via the synapse will fire at a high
rate.
[0048] A voltage dependent connection is different. If the
post-synaptic neuron is already firing at some rate when it
receives a pre-synaptic input signal, then the voltage-dependent
connection will cause the post-synaptic neuron to fire more. Since
the post-synaptic neuron is active, i.e. already firing, this
neuron is at some threshold level. Therefore, the pre-synaptic
connection will modulate the post-synaptic neuron to fire even
more. The voltage-dependent connection, no matter how active the
pre-synaptic neuron is, would have no affect on the post-synaptic
neuron if the latter were not above the threshold value. That is,
the post-synaptic neuron has to have some given threshold of
activity to be responsive or modulated by a voltage-dependent
synaptic connection.
[0049] In the simulated nervous system 12 of FIG. 2, all
within-neural area excitatory connections and all between-neural
area reentrant excitatory connections are voltage-dependent (see
FIG. 2 and Table 2). These voltage-dependent connections, as
described above, play a modulatory role in neuronal dynamics.
[0050] Phase-dependent synaptic connections influence both the
activity, i.e. firing rate, and the phase of post-synaptic neuronal
units, whereas phase-independent synaptic connections influence
only their activity. All synaptic pathways in the simulated nervous
system 12 are phase-dependent except those involved in motor output
(see Table 2: A-left/A-right.fwdarw.C, CC) or sensory input (see
Table 2: Mic-left/Mic-right.fwdarw.A-left/A-right,
A-left.revreaction.A-right, V1.fwdarw.V2), since signals at these
interfaces are defined by magnitude only. Plastic connections are
either value-independent or value-dependent, as described
below.
Neuronal Synchrony in a Simple Network Model
[0051] FIGS. 3A-3E illustrate how reentrant connections among
neuronal units can lead to neuronal synchrony in a mean firing rate
model with a phase parameter as indicated above and, thereby, help
solve the "binding problem" described above. FIG. 3A illustrates a
simple network model consisting of three neuronal units (n1-n3).
Units n1 and n2 receive, respectively, steady phase-independent
input (solid input arrows) and project via respective
voltage-independent connections to the third neuronal unit n3
(solid input arrows). Units n1 and n2 project to each other and
unit n3 projects back to both units n1 and n2, via reentrant
voltage-dependent connections (shown by dotted arrows).
[0052] FIG. 3B is a graph of phase vs. cycle and shows that in this
simplified model all neuronal units n1-n3 become synchronized
within 10 simulation cycles. By contrast, if reentrant connections
are removed (the dotted arrows in FIG. 3A being "lesioned") so that
only feedforward projections remain (the remaining solid arrows in
FIG. 3A), synchrony is not achieved, as shown by the graph of FIG.
3C. While for clarity FIGS. 3B-3C show only the first 15 simulation
cycles, these cycles are representative of network behavior in the
real world of NOMAD 10 over long durations such as 10,000
cycles.
[0053] FIGS. 3D and 3E show the probability distributions from
which postsynaptic phases are chosen for each neuronal unit. With
reentrant connections intact (FIG. 3D), distributions for all
neurons n1-n3 become peaked at the same phase. With reentrant
connections absent, i.e. no reentry ("lesioned" networks, FIG. 3E),
the probability distributions for neuronal units n1 and n2 remain
flat due to their phase-independent inputs, and the distribution
for unit n3 varies randomly over time.
[0054] To explore whether the synaptic property of connection
strength is important for network behavior, the above analysis was
repeated several times using different random seeds, and a network
was compared in which all weights were set to a mean value (1.45).
After 10,000 cycles, qualitatively identical results occurred to
those shown in FIGS. 3B-3E. To explore the effect of the property
of connection plasticity, the above was repeated for networks in
which value-independent plasticity was enabled for the feedforward
projections for neuronal units n1.fwdarw.n3 and n2.fwdarw.n3 (solid
arrows). As before, networks were analyzed with randomly selected
weights as well as networks with all weights set to a mean value
(1.45). In both of these cases, synchrony in intact reentry
networks and no synchrony in lesioned networks occurred. Also,
since pre- and post-synaptic neuronal units were correlated in
activity and phase, plastic connections in the intact networks
increased in strength by nearly 100% over 1000 cycles. In lesioned
networks, however, because pre-and post-synaptic units were not in
phase with each other and these connections were depressed to about
10% of their initial values over the same duration.
[0055] The above indicate the importance of reentry connections to
the "binding problem." That is, the results from this reduced model
of FIG. 3A show that the presence of reentrant connections can
facilitate synchronous activity among neural areas, that this
synchrony does not depend on specific or differential connection
strengths, and that the absence of reentry is not compensated by
synaptic plasticity. The simulated nervous system 12 of the present
invention has three major differences from this reduced model of
FIG. 3A. System 12 has a large-scale reentrant neuroanatomy based
on the vertebrate visual cortex as shown schematically in FIG. 2
and detailed in Table 1 and Table 2 below; it involves
value-dependent and value-independent synaptic plasticity; and it
allows NOMAD 10 to behave autonomously in a real-world
environment.
Neuronal Unit Activity and Phase--Details
[0056] In various embodiments, the mean firing rate (s) of each
neuronal unit ranges continuously from 0 (quiescent) to 1 (maximal
firing). The phase (p) is divided into 32 discrete bins
representing the relative timing of activity of the neuronal units
by an angle ranging from 0 to 2.pi.. The state of a neuronal unit
is updated as a function of its current state and contributions
from voltage-independent, voltage-dependent, and phase-independent
synaptic connectors. The voltage-independent input c to neuronal
unit i from a unit j is:
A.sub.ij.sup.VI(t)=.sub.c.sub.ij.sub.s.sub.j(t),
where s.sub.j(t) is the activity of unit j, and c.sub.ij is the
connection strength from unit j to unit i. The voltage-independent
post-synaptic influence on unit i is calculated by convolving this
value into a cosine-tuning curve over all phases:
POST i VI = l = 1 M j = 1 N l ( A ij VI ( t ) k = 1 32 ( cos ( ( 2
.pi. / 32 ) ( k - p j ( t ) ) ) + 1 2 ) tw ) , ##EQU00001##
where M is the number of different anatomically defined connection
types (see Table 2); N.sub.i is the number of connections of type M
projecting to neuronal unit i; p.sub.j(t) is the phase of neuronal
unit j at time t; and tw is the tuning width, which, in one
embodiment, may be set to 10 so that the width of the tuning curve
is relatively sharp (.about.5 phase bins).
[0057] The voltage-dependent input to neuronal unit i from unit j
is:
A ij VD ( t ) = .PHI. ( POST i VI ( p j ( t ) ) ) c ij s j ( t ) ,
where .PHI. ( x ) = { 0 ; x < .sigma. i vdep x ; otherwise ,
##EQU00002##
where .sigma..sub.i.sup.vdep is a threshold for the post-synaptic
activity below which voltage-dependent connections have no effect
(see Table 1).
[0058] The voltage-dependent post-synaptic influence on unit i is
given by:
POST i VD = l = 1 M j = 1 N l ( A ij VD ( t ) k = 1 32 ( cos ( ( 2
.pi. / 32 ) ( k - p j ( t ) ) ) + 1 2 ) tw ) . ##EQU00003##
[0059] The phase-independent activation into unit i from unit j
is:
A.sub.ij.sup.PI(t)=.sub.c.sub.ijs(t)
[0060] The phase-independent post-synaptic influence on unit i is a
uniform distribution based on all the phase-independent inputs
divided by the number of phase bins (32).
POST i PI ( p ) = l = 1 M j = 1 N l ( A ij PI ( t ) 32 )
##EQU00004##
[0061] A new phase, p.sub.i(t+1), and activity, s.sub.i(t+1) are
chosen based on a distribution created by linearly summing the
post-synaptic influences on neuronal unit i (see FIGS. 4A-4E):
POST i = j = 1 N VI POST j VI + k = 1 N VD POST k VD + l = 1 N PI
POST l PI ##EQU00005##
[0062] The phase threshold, .sigma..sub.i.sup.phase, of the
neuronal unit is subtracted from the distribution POST.sub.i and a
new phase, p.sub.i(t+1), is calculated with a probability
proportional to the resulting distribution (FIG. 4E). If the
resulting distribution has an area less than zero (i.e. no inputs
are above the phase threshold), a new phase, p.sub.i(t+1), is
chosen at random. The new activity for the neuronal unit is the
activity level at the newly chosen phase, which is then subjected
to the following activation function:
si ( t + 1 ) = .phi. ( tanh ( g i ( POST i ( p i ( t + 1 ) ) +
.omega. Si ( t ) ) ) ) , where ##EQU00006## .phi. ( x ) = { 0 ; x
< .sigma. i fire x ; otherwise , ##EQU00006.2##
where .omega. determines the persistence of unit activity from one
cycle to the next, g.sub.i is a scaling factor, and
.sigma..sub.i.sup.fire a unit specific firing threshold.
[0063] Specific parameter values for neuronal units are given in
Table 1, and synaptic connections are specified in Table 2.
TABLE-US-00001 TABLE 1 Neuronal unit parameters. Area Size
.sigma.-fire .sigma.-phase .sigma.-vdep .omega. g V1 (6) 60 .times.
80 -- -- -- -- -- V2 (6) 30 .times. 40 0.10 0.45 0.05 0.30 1.0* V4
(6) 15 .times. 20 0.20 0.45 0.10 0.50 1.0* C 15 .times. 20 0.10
0.10 0.10 0.50 1.0 IT 30 .times. 30 0.20 0.20 0.10 0.75 1.0 S 4
.times. 4 0.10 0.00 0.00 0.15 1.0 Mic-right 1 .times. 1 -- -- -- --
-- Mic-left 1 .times. 1 -- -- -- -- -- A-left 4 .times. 4 0.00 0.00
0.10 0.50 1.0 A-right 4 .times. 4 0.00 0.00 0.10 0.50 1.0
[0064] As shown in Table 1, area V1 is an input neural area and its
activity is set based on the image of camera 16 of FIG. 1. Neural
areas V1, V2 and V4 have six sub-areas each with neuronal units
selective for color (e.g. red and green), and line orientation
(e.g. 0, 45, 90 and 135 degrees). Neural areas Mic-left and
Mic-right are input neural areas and their activity is set based on
inputs from microphones 18, 20 (FIG. 1).
[0065] Table 1 also indicates the number of neuronal units in each
neural area or sub-area ("Size" column). Neuronal units in each
area apart from neural areas V1, Mic-left and Mic-right have a
specific firing threshold (.sigma.-fire), a phase threshold
(.sigma.-phase), a threshold above which voltage-dependent
connections can have an effect (.sigma.-vdep), a persistence
parameter (.omega.), and a scaling factor (g). Asterisks in Table 1
mark values that are set to 1.0 for simulated nervous system 12
(FIG. 2) with lesioned reentrant connections (see Table 2).
TABLE-US-00002 TABLE 2 Properties of anatomical projections and
connection types. Projection Arbor P c.sub.ij(0) type .eta.
.theta..sub.1 .theta..sub.2 k1 k2 V1.fwdarw.V2 .quadrature. 0
.times. 0 1.00 1, 2 PI 0.00 0 0 0.00 0.00 V2.fwdarw.V2(intra)
.quadrature. 3 .times. 3 0.75 0.45, 0.85 VD 0.00 0 0 0.00 0.00
V2.fwdarw.V2(inter) (X) .quadrature. 2 .times. 2 0.40 0.5, 0.65 VD
0.00 0 0 0.00 0.00 V2.fwdarw.V2(intra) .crclbar. 18, 25 0.10 -0.05,
-0.1 VI 0.00 0 0 0.00 0.00 V2.fwdarw.V2(inter) .quadrature. 2
.times. 2 0.05 -0.05, -0.1 VI 0.00 0 0 0.00 0.00 V2.fwdarw.V4
.quadrature. 3 .times. 3 0.40 0.1, 0.12 VI 0.00 0 0 0.00 0.00
V4.fwdarw.V2 (X) .quadrature. 1 .times. 1 0.10 0.25, 0.5 VD 0.00 0
0 0.00 0.00 V4.fwdarw.V4(inter) (X) .quadrature. 2 .times. 2 0.40
1.75, 2.75 VD 0.00 0 0 0.00 0.00 V4.fwdarw.V4(intra) .crclbar. 10,
15 0.10 -0.15, -0.25 VI 0.00 0 0 0.00 0.00 V4.fwdarw.V4(inter)
.crclbar. 10, 15 0.10 -0.15, -0.25 VI 0.00 0 0 0.00 0.00
V4.fwdarw.V4(inter) .quadrature. 2 .times. 2 0.03 -0.15, -0.25 VI
0.00 0 0 0.00 0.00 V4.fwdarw.C .quadrature. 3 .times. 3 1.00 0.002,
0.0025 VI 0.00 0 0 0.00 0.00 V4.fwdarw.IT special -- 0.1, 0.15 VI
0.00 0 0 0.00 0.00 IT.fwdarw.V4 (X) non-topo 0.01 0.05, 0.07 VD
0.00 0 0 0.00 0.00 IT.fwdarw.IT non-topo 0.10 0.14, 0.15 VD 0.10 0
0.866 0.90 0.45 IT.fwdarw.C # non-topo 0.10 0.2, 0.2 VD 1.00 0
0.707 0.45 0.65 IT.fwdarw.S # non-topo 1.00 0.0005, 0.001 VI 0.10 0
0.707 0.45 0.45 C.fwdarw.V4 (X) non-topo 0.01 0.05, 0.07 VD 0.00 0
0 0.00 0.00 C.fwdarw.C .crclbar. 6, 12 0.50 -0.05, -0.15 PI 0.00 0
0 0.00 0.00 C.fwdarw.Mleft non-topo 1.00 35, 35 VD 0.00 0 0 0.00
0.00 C.fwdarw.Mright non-topo 1.00 35, 35 VD 0.00 0 0 0.00 0.00
S.fwdarw.C non-topo 0.50 0.5, 05 VD 0.00 0 0 0.00 0.00 S.fwdarw.S
non-topo 0.50 0.7, 0.8 VD 0.00 0 0 0.00 0.00 A-left.fwdarw.C
left-only 1.00 0.5, 0.5 VD 0.00 0 0 0.00 0.00 A-right.fwdarw.C
right-only 1.00 0.5, 0.5 VD 0.00 0 0 0.00 0.00 A-left.fwdarw.C
right-only 1.00 -0.15, -0.15 PI 0.00 0 0 0.00 0.00 A-right.fwdarw.C
left-only 1.00 -0.15, -0.15 PI 0.00 0 0 0.00 0.00 A-left.fwdarw.S
non-topo 1.00 35, 35 VD 0.00 0 0 0.00 0.00 A-right.fwdarw.S
non-topo 1.00 35, 35 VD 0.00 0 0 0.00 0.00 A-left A-right non-topo
1.00 -1, -1 PI 0.00 0 0 0.00 0.00 A-left A-right non-topo 1.00
-0.5, -0.5 VD 0.00 0 0 0.00 0.00 Mic-left, Mic-right.fwdarw.A-left,
A-right non-topo 1.00 5, 5 PI 0.00 0 0 0.00 0.00
[0066] Table 2 shows properties of anatomical projections and
connection types of simulated nervous system 12. A pre-synaptic
neuronal unit connects to a post-synaptic neuronal unit with a
given probability (P) and given projection shape (Arbor). This
arborization shape can be rectangular ".quadrature." with a height
and width (h.times.w), doughnut shaped ".THETA." with the shape
constrained by an inner and outer radius (r1, r2), left-only
(right-only) with the pre-synaptic neuronal unit only projecting to
the left (right) side of the post-synaptic area, or
non-topographical ("non-topo") where any pairs of pre-synaptic and
post-synaptic neuronal units have a given probability of being
connected. The initial connection strengths, C.sub.ij(O), are set
randomly within the range given by a minimum and maximum value
(min, max). A negative value for C.sub.ij(O), indicates inhibitory
connections. Connections marked with "intra" denote those within a
visual sub-area and connections marked with "inter" denote those
between visual sub-areas. Inhibitory "inter" projections connect
visual sub-areas responding to shape only or to color only (e.g.
V4-redV4-green, V4-horizontalV4-vertical), excitatory "inter"
projections connect shape sub-areas to color sub-areas (e.g.
V4-redV4-vertical). Projections marked # are value-dependent. A
connection type can be phase-independent/voltage-independent (PI),
phase-dependent/voltage-independent (VI), or
phase-dependent/voltage-dependent (VD). Non-zero values for .eta.,
.theta..sub.1, .theta..sub.2, k.sub.1, and k.sub.2 signify plastic
connections. The connection from V4 to IT was special in that a
given neuronal unit in area IT was connected to three neuronal
units randomly chosen from three different V4 sub-areas.
Projections marked with an "X" were removed during lesion
experiments.
[0067] In this model of a neuronal unit, post-synaptic phase tends
to be correlated with the phase of the most strongly active
pre-synaptic inputs. This neuronal unit model facilitates the
emergence of synchronously active neuronal circuits in both a
simple network (see FIG. 3A above, Neuronal Synchrony in a Simple
Network Model) and in the full simulated nervous system (FIG. 2),
where such emergence involves additional constraints imposed by
reentrant connectivity, plasticity, and behavior.
Synaptic Plasticity
[0068] Synaptic strengths are subject to modification according to
a synaptic rule that depends on the phase and activities of the
pre- and post-synaptic neuronal units. Plastic synaptic connections
are either value-independent (see IT.fwdarw.IT in FIG. 2) or
value-dependent (see IT.fwdarw.S, IT.fwdarw.C in FIG. 2). Both of
these rules are based on a modified BCM learning rule in which
thresholds defining the regions of depression and potentiation are
a function of the phase difference between the pre-synaptic and
post-synaptic neuronal units (see FIG. 2, inset). The graphical
inset shown in FIG. 2 shows a form of the known BCM rule in which
synaptic change (.DELTA.C.sub.ij) is a function of the phase
difference between post-and pre-synaptic neuronal units (.DELTA.P)
and two thresholds (.theta..sub.1 and .theta..sub.2).
[0069] Synapses between neuronal units with strongly correlated
firing phases are potentiated and synapses between neuronal units
with weakly correlated phases are depressed; the magnitude of
change is determined as well by pre- and post-synaptic activities.
This learning rule is similar to a spike-time dependent plasticity
rule applied to jittered spike trains where the region of
potentiation has a high peak and a thin tail, and the region of
depression has a comparatively small peak and fat tail.
[0070] Value-independent synaptic changes in c.sub.ij are given
by:
.DELTA..sub.c.sub.ij(t+1)=.eta..sub.s.sub.i(t).sub.s.sub.j(t)BCM(.DELTA.-
p),
where s.sub.i(t) and s.sub.j(t) are activities of post- and
pre-synaptic units, respectively, .eta. is a fixed learning rate,
and
.DELTA. p = cos ( ( 2 .pi. / 32 ) ( p i ( t ) - p j ( t ) ) ) + 1 2
, ##EQU00007##
where p.sub.i(t) and p.sub.j(t) are the phases of post- and
pre-synaptic units (0.0.ltoreq..DELTA.p.ltoreq.1.0). A value of
.DELTA.p near 1.0 indicates that pre-and post-synaptic units have
similar phases, a value of .DELTA.p near 0.0 indicates that pre-
and post-synaptic units are out of phase. The function BCM is
implemented as a piecewise linear function, taking .DELTA.p as
input, that is defined by two thresholds (.theta..sub.1,
.theta..sub.2, in radians), two inclinations (k.sub.1, k.sub.2) and
a saturation parameter .rho. (.rho.=6 throughout):
BCM ( .DELTA. p ) = { 0 ; .DELTA. p < .theta. 1 k 1 ( .theta. 1
- .DELTA. p ) ; .theta. 1 .ltoreq. .DELTA. p < ( .theta. 1 +
.theta. 2 ) / 2 k 1 ( .DELTA. p - .theta. 2 ) ; ( .theta. 1 +
.theta. 2 ) / 2 .ltoreq. .DELTA. p < .theta. 2 k 2 tanh ( .rho.
( .DELTA. p - .theta. 2 ) ) / .rho. ; otherwise , ##EQU00008##
[0071] Specific parameter settings for fine-scale synaptic
confections are given in Table 2.
[0072] The rule for value-dependent synaptic plasticity differs
from the value-independent rule in that an additional term, based
on the activity and phase of the value system (neural areas),
modulates the synaptic strength changes. Synaptic connections
terminating on neuronal units that are in phase with the value
system are potentiated, and connections terminating on units out of
phase with the value system are depressed.
[0073] The synaptic change for value-dependent synaptic plasticity
is given by:
.DELTA..sub.c.sub.ij(t+1)=.eta..sub.s.sub.i(t).sub.s.sub.j(t)BCM(.DELTA.-
p)V(t)BCM.sub.v(.DELTA.p.sub.v),
where V(t) is the mean activity level in the value areas S at time
t. Note that the BCM.sub.v function is slightly different than the
BCM function above in that it uses the phase difference between
area S and the post-synaptic neuronal unit as input
( .DELTA. p v = cos ( ( 2 .pi. / 32 ) ( p V ( t ) - p i ( t ) ) ) +
1 2 , ##EQU00009##
where p.sub.v(t) is the mean phase in area S. When both BCM and
BCM.sub.v return a negative number, BCM.sub.v is set to 1 to ensure
that the synaptic connection is not potentiated when both the
pre-synaptic neuronal unit and value system (neural areas) are out
of phase with the post-synaptic neuronal unit.
Simulated Cycle Computation
[0074] During each simulation cycle of simulated nervous system 12,
sensory input is processed, the states of all neuronal units are
computed, the connection strengths of all plastic connections are
determined, and motor output is generated. In experiments described
below, execution of each simulated cycle required approximately 100
milliseconds of real time.
Experimental Conditions
[0075] FIG. 5A shows a diagram of the environment of NOMAD 10. The
environment consisted of an enclosed area with black walls. Various
pairs of shapes from a set consisting of a green diamond, a green
square, a red diamond, and a red square were hung on two opposite
walls. The floor was covered with opaque black plastic panels, and
contained a boundary made of reflective construction paper. When
this boundary was detected by the infrared (IR) detector attached
to the front of NOMAD 12 and facing toward the floor, NOMAD 10 made
one of two reflexive movements: (i) if an object was in its visual
field, it backed up, stopped and then turned roughly 180 degrees,
(ii) if there was no object in its visual field, NOMAD 10 turned
roughly 90 degrees, thus orienting away from walls without visual
stimuli. Near the boundary of walls containing visual shapes,
infrared emitters (IR) on one side of the room were paired with IR
sensors containing speakers on the other side (as shown in FIG.
5A), to create an IR beam. If the movement of NOMAD 10 broke either
IR beam, a tone was emitted by the speakers. Detection of the tone
by NOMAD 10 elicited an orientation movement towards the source of
the sound via the simulated nervous system 12.
Experimental Protocol--FIGS. 6A-6B
[0076] FIGS. 6A and 6B illustrate the experimental set-up for NOMAD
10. NOMAD 10 views objects on two of the walls of an area, which is
about "90 by 66". Experiments were divided into two stages,
training and testing, as shown in FIGS. 6A and 6B, respectively.
During both stages the activity and phase responses of all neuronal
units of neural areas V1, V2, etc. were recorded for analysis.
[0077] During training as shown in FIG. 6A, NOMAD 10 explored its
enclosure for 10,000 simulation cycles corresponding to roughly 24
approaches to the pairs of various objects shown. Responses to
sounds emitted by the speaker (auditory cues) caused NOMAD 10 to
orient toward the target, which in this example is the red diamond.
The distracters, which were exchanged before every sixth approach
to ensure that left-right orientation of NOMAD 10 did not confound
target relation, are a green diamond and a red square. For testing,
as shown in FIG. 6B, the speakers were turned off and NOMAD 10 was
allowed to explore the environment for 15,000 simulation cycles.
While the target object was continuously present for these 15,000
cycles, the distracters were changed every 5,000 cycles.
Training Stage Details--FIG. 6A
[0078] In the training stage shown in FIG. 6A, NOMAD 10
autonomously explored its enclosure for 10,000 simulation cycles,
corresponding to 15-20 minutes of real time and approximately 24
approaches to the various pairs of visual shapes which were a red
diamond and red square (on the left of FIG. 6A) and a red diamond
and a green diamond (on the right of FIG. 6A). Thus, each pair
contained a "target" shape (red diamond) and a "distracter" shape
(green diamond on a red square). Distracters were deliberately
designed to share attributes with the target, for example, when the
red diamond was the target, a red diamond/red square pair was hung
on one wall (shown on left side of FIG. 6A), and a red
diamond/green diamond pair was hung on the other wall (shown on
right side of FIG. 6A). The red diamond on either side of the room
was closest to the speakers in both cases, as illustrated. To
ensure that the left-right orientation of shapes in the
target-distracter pair (e.g. red-square on the left, green-diamond
on the right) did not confound target selection, the side of the
distracters were exchanged every sixth viewing of a pair. During
the training stage, responses to the speakers caused NOMAD 10 to
orient towards the target.
Testing Stage Details--FIG. 6B
[0079] During testing, as shown in FIG. 6B, the speakers were
turned off (therefore not shown), and NOMAD 10 was allowed to
autonomously explore its enclosure for 15,000 simulation cycles.
The first 10,000 cycles involved encounters with the same target
and distracters present during the training stage of FIG. 6A. The
final 5,000 cycles involved encounters with the target and the
single shape of the set of four shapes (left and right) that did
not share any features with the target (e.g. a pair consisting of a
red diamond as target and a green square as distracter).
[0080] Training and testing were repeated with three different
"subjects" of the brain-based device BBD using each of the four
shapes as a target (a total of 12 training and testing sessions).
Each BBD "subject" had the same physical device of NOMAD 10, but
each possessed a unique simulated nervous system 24. This
variability among "subjects" was a consequence of random
initialization in both the microscopic details of connectivity
between individual neuronal units and the initial connection
strengths between those neuronal units. The overall connectivity
among neuronal units remained similar among different "subjects",
however, inasmuch as that connectivity was constrained by the
synaptic pathways, arborization patterns, and ranges of initial
connection strengths (see FIG. 2 and Table 2 for specifics).
Target Tracking Behavior--Generally--FIGS. 7A-7B
[0081] The discrimination performance of each "subject" of the
brain-based device BBD was assessed by how well that "subject"
tracked toward target objects in the absence of auditory cues
following conditioning or training, as shown in FIGS. 7A-7B. This
was calculated as the fraction of time for which the target was
centered in NOMAD 10's visual field via camera 16 during each
approach to a pair of visual objects shown in FIG. 6B. Three
separate "subjects" were conditioned to prefer one of four target
shapes or objects, i.e. red diamond (rd), red square (rs), green
square (gs) and green diamond (d). Activity in neural area V2 was
used to assess the percentage of time for which the visual field of
NOMAD 10 via its camera 16 was centered on a particular visual
shape. Bars in the graphs of FIGS. 7A and 7B represent the mean
percentage tracking time with error bars denoting the standard
deviation. As shown in FIG. 7A, BBD "subjects" with intact
reentrant connectors tracked the targets (white bars) significantly
more than the distracters (gray bars) for each target shape,
averaging over all approaches (* denote p<0.01 using a paired
sample nonparameter sign test). As shown in FIG. 7B, "subjects"
with reentrant connections intact (white bars) tracked targets
significantly better than "subjects" with "lesions" only during
testing (light gray bars), and subjects with lesions during both
training and testing (black bars) (* denote p<0.01 using a
RankSum test).
[0082] FIG. 7A shows that all "subjects" successfully tracked the
four different targets over 80% of the time. This, despite the fact
that the targets and distracters appeared in the visual field of
camera 16 at many different scales and at many different positions
as NOMAD 10 explored its environment (invariant object recognition
described below). Moreover, NOMAD 10 achieved this process even
though because of shared properties (e.g. same color or same
shape), targets cannot be reliably distinguished from distracters
on the basis of color or shape alone.
[0083] To investigate the importance of the presence of reentrant
connections in the various "subjects" of the brain-based device
BBD, certain inter-areal reentrant connections were lesioned at
different stages of the experimental paradigm with the results
shown in FIG. 7B. In one case, previously trained "subjects" were
retested after lesioning. In a second case, reentrant connections
were lesioned in both training and testing stages. Lesions were
applied to a subset of inter-areal excitatory reentrant connections
(see projections marked with an "X" in FIG. 2 and in Table 2),
which had the effect of transforming the simulated nervous system
12 into a "feed-forward" model of visual processing. To compensate
for the reduction in activity due to these lesions, neuronal unit
outputs in areas V2 and V4 were amplified (see Table 1). FIG. 7B
shows that "subjects" with intact reentrant connections performed
significantly better than either lesioned group. The decrease in
performance observed in the absence of reentry connectors indicates
that reentrant connections are essential for behavior, above
chance, in the object discrimination task.
Neural Dynamics During Behavior--FIG. 8
[0084] During the behavior of NOMAD 10 in its environment, circuits
comprised of synchronously active neuronal groups were distributed
throughout different neural areas in the simulated nervous system
12. Multiple objects in the environment were distinguishable by the
differences in phase between the corresponding active circuits. A
snapshot of the neural responses during a typical behavioral run is
given in FIG. 8. This snapshot shows NOMAD 10 during an approach to
a red diamond target and a green diamond distracter towards the end
of a training session (FIG. 6A). Each pixel in the depicted neural
areas V2, V4, IT, C and S represents the activity and phase of a
single neuronal unit within the respective given neural area. Thus,
for example, FIG. 8 shows the responses for neural areas V2 and V4
specifically their neural sub-areas in color (red, green) and line
orientation (vertical, diagonal). The phase is indicated by the
color of each pixel and the activity is indicated by brightness of
the pixel (black is no activity; very bright is maximum
activity).
[0085] FIG. 8 shows two neural circuits which are differentiated by
their distinct phases and which were elicited respectively by the
red diamond and the green diamond stimuli. As shown in the figure,
NOMAD 10 has not yet reached the IR beam that triggers the speakers
in its environment to emit a tone (see FIG. 6A). The activity of
neural area S (the value system) was nonetheless in phase with the
activity in neural areas V2 and V4 corresponding to the target, and
was therefore predictive of the target's saliency or value. Area IT
has two patterns of activity, indicated by the two different phase
colors, which reflect two perceptual categories. These patterns
were brought about by visual input from camera 16 that is generated
during the movement of NOMAD 10 in this environment. Finally,
neural area C has more activity on the side that facilitates
orientation of NOMAD 10 towards the target (i.e. the red
diamond).
Dynamics of Neural Responses--FIGS. 9A and 9B
[0086] To analyze the dynamics of these neural responses, the phase
distributions of active neuronal units during approaches to
target-distracter pairs in the testing sessions were examined. FIG.
9A shows the distribution of neuronal phases in various neural
areas during approaches to a red diamond target in the presence of
a red square distracter, by an intact "subject" (with reentrant
connections). FIG. 9A shows consistent correlations among phase
distributions in neural sub-areas V4R (red), V4H (horizontal), and
V4D (diagonal). The bimodal distribution in neural sub-area V4R
reflects the presence of two red shapes (diamond, square) in the
environment of NOMAD 10: one trace correlates reliably with neural
sub-area V4D (diagonal) and can therefore be associated with the
red diamond, the other correlates reliably with sub-area V4H
(horizontal) and can be associated with the red square (not shown
is area V4G which remained inactive during this period). The phases
of the active neuronal units in areas S, IT, and C were strongly
correlated with the red diamond target, as opposed to the red
square distracter, reflecting the synaptic changes brought about by
previous conditioning during the testing phase to prefer the red
diamond. The global pattern of network activity thus displayed a
biased phase distribution in favor of the target.
Quantification of Biased Phase Distribution--FIGS. 9A-9B; Table
3
[0087] To quantify this bias and assess its generality, the
proportion of neuronal units in areas S, IT, and C associated with
the target with the proportion associated with the distracter
during the testing. Table 3 shows average values of these
proportions calculated over all "subjects" and all four target
shapes.
TABLE-US-00003 TABLE 3 Neuronal composition and average activity of
functional circuits corresponding to target and distracter objects
mean firing mean firing % units % units rate of units rate of units
responding responding responding responding Area to target to
distracter to target to distracter S 61.25 (17.78)* 10.34 (4.88)
0.537 (0.089) 0.495 (0.080) IT 4.29 (0.495)* 2.91 (0.366) 0.579
(0.064)* 0.467 (0.015) C 19.04 (1.45)* 10.80 (1.93) 0.626 (0.062)*
0.398 (0.032) V4 14.83 (0.833)* 11.61 (0.819) 0.829 (0.003)* 0.823
(0.002)
A significantly greater proportion of neuronal units were part of
functional circuits associated with targets than in circuits
associated with distracters. In addition, those neuronal units
associated with targets had significantly higher firing rates than
neuronal units in circuits associated with distracters.
[0088] The above shows that perceptual categorization and visual
object discrimination by NOMAD 10 is enabled by the coherent
interaction of local and global neuronal circuit processes, as
mediated by reentrant connections, of simulated nervous system 12.
Local processes correspond to activity in each neural area, whereas
global processes correspond to the distinct, but distributed
functional circuits that emerged throughout the simulated nervous
system 12. These interactions are evident in FIG. 9A, in which
activity in each of the local areas strongly reflects the global
bias in favor of the red-diamond target (see also Table 3).
The Influence of Reentry on Neural Dynamics
[0089] Lesioning of reentrant connections interfered significantly
with interactions between the local and global processes mentioned
above. Even in a very simple network model, removal of reentrant
connections can prevent the emergence of neural synchrony (see
FIGS. 3A-3E). On a larger scale, FIG. 9B shows approaches by the
same NOMAD 10 "subject" depicted in FIG. 9A to the same
target/distracter pair, following lesions of inter-areal excitatory
reentrant connections. While some individual areas continued to
show peaks in their phase distribution (e.g. neural sub-area V4R),
many do not, and the phase correlations between the neural areas
are severely diminished. This occurred not only among the various
V4 neural areas (FIG. 2), but also among area V4 and areas S, IT,
and C. The dynamically formed and globally coherent circuits, which
were clearly evident in the intact "subject", were almost entirely
absent in the lesioned "subjects". For example, FIG. 9B shows that
activity in area S no longer correlates uniquely with a single
trace in area V4; instead, it alternates between two distinct
states. The absence of a dominant trace in neural areas IT and C is
also shown.
[0090] Phase correlations between neural areas were significantly
higher for "subjects" with intact reentrant connections than for
"subjects" in either lesion group. The overall median rank
correlation coefficient was 0.36 for the intact "subjects", 0.21
for the "subjects" with lesions only during the test stage, and
0.17 for the "subjects" with lesions in both the training and test
stages. Also, "subjects" with lesions only during testing had
significantly higher correlation coefficients than "subjects" with
lesions during both training and testing. This reflects the
contribution of reentrant connections to the formation of global
circuits during training (FIG. 6A). All of these findings are
consistent with the drop in behavioral performance in the absence
of reentrant connections (see FIG. 7).
Phase Correlations Among Neural Areas--Single "Subject" Conditioned
to a Red Diamond Shape
[0091] FIGS. 10A-10C illustrate geographically a representative
example of the correlation of phases among neural areas for a
"subject" after conditioning to prefer red diamond targets. The
figures are color coded (dark blue denotes no correlation, dark red
denotes high correlation), and each colored area shows the
correlation coefficient between the mean phases of a given pair of
neural areas. FIG. 10A shows correlation coefficients when
reentrant connections were intact. In agreement with the data shown
in FIG. 9, strong phase correlations were found between areas
associated with specific target features (V4D and V4R), and among
these areas and areas S, IT and C. The correlations among neural
areas for the same "subject" with reentrant connections lesioned
during testing (FIG. 10B) and with reentrant connections lesioned
during both conditioning and testing (FIG. 10C) were both
considerably weaker. As graphically indicated, the connections
between neural areas V1, V2, etc. associated with the target are
considerably higher with reentrant connections intact (FIG. 10A)
than in either lesion case (FIGS. 10B and 10C).
Invariant Object Recognition--FIGS. 11A and 11B
[0092] FIGS. 11A and 11B are graphs illustrating the response of
neural value area S to target objects in the visual field of NOMAD
10 at different positions and at different scales. Average values
were calculated for all approaches by NOMAD 10 "subjects" to all
target objects, with the error bars indicating standard errors.
FIG. 11A shows average responses as a function of target position
within the visual field (135.degree.). FIG. 11B shows average
responses as the apparent target size ranged from 8.degree. to
27.degree. of visual angle. The insets in FIGS. 11A and 11B
indicate how the square target appears in NOMAD 10's field of view
at extreme positions and scales.
[0093] Because images of the visual objects varied considerably in
size and position as NOMAD 10 explored its enclosure, successful
discrimination required invariant object recognition. In order to
analyze this capacity, the value system, i.e. neural area S, was
examined which, after conditioning, responded preferentially to
target objects over distracters due to plasticity in the pathway
IT.fwdarw.S. In a typical approach, as NOMAD 10 moved from one side
of the environment to the other, neural area S responded briskly
and in phase with neuronal units in areas V2, V4, and IT
corresponding to attributes of the target. Calculating average
values over all "subjects" and all target shapes, it was found that
area S responded reliably to target images which appeared within
120.degree. of the center of the field of view (the range of the
visual field was approximately .+-.35.degree.) and as the apparent
target size ranged from 8.degree. to 27.degree. of visual angle.
Thus, the object recognition of the brain-based device BBD of the
present invention while autonomously moving in its environment was
both position and scale invariant.
Value System (Neural Area S) Activity During Conditioning--FIGS.
12A and 12B
[0094] Neural activity during conditioning for a single "subject",
for neural areas S, IT, and C during a single approach to a target
shape is shown in FIG. 12A, at an early stage (left panels, time
steps 750-1165), and in FIG. 12B, at a late stage of conditioning
(right panels, time steps 6775-7170). Each panel shows the
distribution of neuronal unit phases in the corresponding neural
area over time. As in FIGS. 9A and 9B, a gray scale indicates the
proportion of neuronal units in each neural area at a particular
phase. The solid line at the bottom of each panel indicates time
steps for which the tone from a speaker was present (see FIG. 6A).
In the early conditioning training period (left panels) area S is
inactive until tone onset, i.e. an audible activity, at which point
it becomes strongly activated in phase with the upper traces in
both areas IT and C, which are associated with the target. The
lower traces in areas IT and C, corresponding to the distracter,
become relatively suppressed at the same time. Later in
conditioning (right panels), areas S, IT and C are in phase with
visual system activity corresponding to the target (lower trace)
well before tone onset, and activity associated with the distracter
is relatively suppressed well before the tone onset.
[0095] As a result of value-dependent synaptic plasticity during
conditioning (i.e. the plasticity of the synaptic connectors are
dependent on value), the visual attributes of target objects became
predictive of value. As shown in FIG. 12A, during early
conditioning area S does not become active until the UCS
(unconditioned stimulus; i.e. the tone) is present. The UCS also
evokes biases in areas IT and C, as shown by the rapid abolition of
the initially bimodal phase distributions in these areas.
[0096] At a later stage of conditioning, the CS (the conditioned
stimulus; i.e. the target visual features) has become associated
with value such that activity in area S now precedes UCS onset (see
FIG. 12B). Area S responds to the target stimulus as soon as the
stimulus appears in NOMAD 10's visual field. Activity in area S
then facilitates a bias in areas IT and C, as shown in FIGS.
12A-12B by the appearance of a single phase distribution peak in
each area well before UCS onset. This shift in the timing of
value-related activity, from activity triggered by the auditory UCS
in early trials (i.e. auditory input provides value), to activity
triggered by the visual CS in later trials (i.e. visual input now
provides value), is analogous to the shift in dopaminergic neural
activity found in the primate ventral tegmental area during
conditioning. Value-dependent synaptic plasticity is also similar
to "temporal-difference" learning in that the conditioned stimulus
becomes predictive of value.
Computer System and Flow Charts
[0097] FIG. 13 is an exemplary illustration of a system in
accordance with various embodiments of the invention. Although this
diagram depicts components as logically separate, such depiction is
merely for illustrative purposes. It will be apparent to those
skilled in the art that the components portrayed in this figure can
be arbitrarily combined or divided into separate software, firmware
and/or hardware components. Furthermore, it will also be apparent
to those skilled in the art that such components, regardless of how
they are combined or divided, can execute on the same computing
device or can be distributed among different computing devices
connected by one or more networks or other suitable communication
means.
[0098] In various embodiments, the components illustrated in FIG.
13 can be implemented in one or more programming languages (e.g.,
C, C++, Java.TM., and other suitable languages). Components can
communicate using Message Passing Interface (MPI) or other suitable
communication means, including but not limited to shared memory,
distributed objects and Simple Object Access Protocol (SOAP). MPI
is an industry standard protocol for communicating information
between computing devices (or nodes). In one embodiment, the system
can be deployed on a multi-processor computer architecture such as
(but not limited to) a Beowulf cluster. Beowulf clusters are
typically comprised of commodity hardware components (e.g.,
personal computers running the Linux operating system) connected
via Ethernet or some other network. The present disclosure is not
limited to any particular type of parallel computing architecture.
Many other such architectures are possible and fully within the
scope and spirit of the present disclosure.
[0099] Referring to FIG. 13, master component 1302 can coordinate
the activities of the other components according to commands
received from client 1304. In one embodiment, the client can be a
stand-alone process that programmatically controls the master
according to a script or other scenario and/or in reaction to
client information (e.g., neural activity, sensor readings and
camera input) received from the master. Client commands can
instruct the master to start or stop the brain-based device BBD
experiment, save the experiment state on data store 1312, read the
experiment state from the data store, set the running time/cycles
in which the experiment will execute, and set parameters of the
neural simulators 1310.
[0100] In another embodiment, the client can be a user interface
that receives information from the master and allows a user to
interactively control the system. By way of a non-limiting example,
a user interface can include one or more of the following: 1) a
graphical user interface (GUI) (e.g., rendered with Hypertext
Markup Language); 2) an ability to respond to sounds and/or voice
commands; 3) an ability to respond to input from a remote control
device (e.g., a cellular telephone, a PDA, or other suitable remote
control); 4) an ability to respond to gestures (e.g., facial and
otherwise); 5) an ability to respond to commands from a process on
the same or another computing device; and 6) an ability to respond
to input from a computer mouse and/or keyboard. This disclosure is
not limited to any particular UI. Those of skill in the art will
recognize that many other user interfaces are possible and fully
within the scope and spirit of this disclosure.
[0101] The neuronal units for each neural area (e.g., V1, V2, V4,
IT, C, S, Mic-left, A-left, Mic-right, A-right) are each assigned
to a neural simulator 1310. Each neural simulator 1310 is
responsible for calculating the activity of the neuronal units that
have been assigned to it. A given neural area's neuronal units may
be distributed across one or more neural simulators 1310. In
various embodiments, there can be one neural simulator per Beowulf
node. In order to optimize performance, neuronal units can be
distributed among neural simulators such that the average number of
synaptic connections on the neural simulators is approximately the
same. In other embodiments, neuronal units can be distributed such
that the average number of neuronal units per neural simulator is
approximately the same. Neural simulators periodically or
continuously exchange the results of calculating the activity of
their neuronal units with other neural simulators and the master.
This information is required so that neuronal units on other neural
simulators have up-to-date pre-synaptic inputs. The master provides
actuator commands to the NOMAD based on the neural activity
received from the neural simulators.
[0102] The master periodically receives image data from image
grabber 1306 and distributes it to the neural simulators and to the
client. In one embodiment, the images are taken from the CCD camera
16 mounted on NOMAD 10 that sends 320.times.240 pixel RGB video
images, via an RF transmitter, to an ImageNation PXC200 frame
grabber. The image is then spatially averaged to produce an
80.times.60 pixel image. Gabor filters can be used to detect edges
of vertical, horizontal, and diagonal (45 and 135 degrees)
orientations (as briefly described above). The output of the Gabor
function is mapped directly onto the neuronal units of the
corresponding V1 sub-area. Color filters (red positive center with
a green negative surround, or red negative center with a green
positive surround) are also applied to the image. The outputs of
the color filters are mapped directly onto the neuronal units of
V1-Red and V1-Green. V1 neuronal units projected retinotopically to
neuronal units in neural area V2.
[0103] The master component also periodically acquires sensor data
from NOMAD 10 component 1308 and distributes it to the neural
simulators. In one embodiment, a micro controller (PIC17C756A)
onboard the NOMAD 10 samples input and status from its sensors and
controls an RS-232 communication between the NOMAD base and master.
Sensor information can include, in addition to video and audio
information previously described, gripper state, camera position,
infrared detectors, whisker deflection, wheel speed and direction,
odometer count, and microphone input. In one embodiment, a root
mean square (RMS) chip measures the amplitude of the microphone
input signal and a comparator chip produces a square waveform which
allows frequency to be measured. A micro controller on NOMAD 10
periodically calculates the overall microphone amplitude by
averaging the current signal amplitude measurement with the
previous three measurements. The micro controller calculates the
frequency of the microphone signal at each time point by inverting
the average period of the last eight square waves. Neural areas
Mic-left and Mic-right respond to tones between 2.9 and 3.5 kHz
having an amplitude of at least 40% of the maximum. The activity of
a neuronal unit in neural area Mic-left or Mic-right is given
by
s.sub.i.sup.mic(t+1)=tan
h(0.9s.sub.i.sup.mic(t)+0.1a.sub.i.sup.mic),
where s.sub.i.sup.mic(t) is the previous value of a neuronal unit i
in Mic-left or Mic-right, and a.sub.i.sup.mic is the current
amplitude of the microphone output.
[0104] FIG. 14 is a flow diagram illustration of neural simulator
initialization in accordance with various embodiments of the
invention. Although this figure depicts functional steps in a
particular order for purposes of illustration, the process is not
necessarily limited to any particular order or arrangement of
steps. One skilled in the art will appreciate that the various
steps portrayed in this figure can be omitted, rearranged,
performed in parallel, combined and/or adapted in various ways. In
step 1402, it is determined based on command(s) from the client
1304 whether or not a saved experiment should be retrieved from the
data store 1312 or whether a new experiment should be started. If
the experiment is to be retrieved from the data store, this is
performed in step 1410. In various embodiments, the experiment
state can be stored as an Extensible Markup Language (XML)
document, a plain text file, or a binary file. Otherwise, in step
1404 neuronal units are created according to the parameters given
in Table 1. Next, in step 1406 synaptic connections are created
between the neuronal units according to the parameters in Table 2.
Finally, each neuronal unit is assigned to a neural simulator in
step 1408.
[0105] FIG. 15 is a flow diagram illustration of the master
component in accordance with various embodiments of the invention.
Although this figure depicts functional steps in a particular order
for purposes of illustration, the process is not necessarily
limited to any particular order or arrangement of steps. One
skilled in the art will appreciate that the various steps portrayed
in this figure can be omitted, rearranged, performed in parallel,
combined and/or adapted in various ways.
[0106] In step 1502 the master broadcasts image and sensor data
that it has acquired from the image grabber and NOMAD 10 to the
neural simulators and the client. In step 1504, the master
broadcasts any commands it may have received to the neural
simulators. In step 1506, it is determined whether or not the
client has directed the master to quit the experiment. If so, the
master ceases the experiment (which may include saving the state of
the experiment to the data store). Otherwise, in step 1508 the
updated information is provided to the client which could serve to
update a GUI. In step 1510, neuronal unit activity from the neural
simulators is shared among all components (e.g., via MPI). The
neuronal activity can be provided in some form to the client as
part of the client information. Finally, it is determined whether
or not there are any remaining cycles left in the simulation. If
not, the experiment terminates. Otherwise, the master returns to
step 1502.
[0107] FIG. 16 is a flow diagram illustration of a neural simulator
in accordance to various embodiments of the invention. Although
this figure depicts functional steps in a particular order for
purposes of illustration, the process is not necessarily limited to
any particular order or arrangement of steps. One skilled in the
art will appreciate that the various steps portrayed in this figure
can be omitted, rearranged, performed in parallel, combined and/or
adapted in various ways.
[0108] In step 1602, the neural simulator accepts image and sensor
data that is broadcast by the master. In step 1604, client commands
broadcast by the master are accepted. In step 1606, it is
determined whether or not the client has directed the master to
quit the experiment. If so, the neural simulator completes its
execution. Otherwise, in step 1608 the value of the neuronal units
assigned the neural simulator are calculated. In step 1610, the
strengths of plastic connections are calculated. Local neuronal
unit activity is shared in step 1612 with other neural simulators
and the master. In addition, neuronal activity from other neural
simulators is acquired and used to refresh local values. Finally,
it is determined in step 1614 whether or not there are any
remaining cycles left in the simulation. If not, the experiment
terminates. Otherwise, the neural simulator returns to step
1602.
[0109] Various embodiments may be implemented using a conventional
general purpose or a specialized digital computer or
microprocessor(s) programmed according to the teachings of the
present disclosure, as will be apparent to those skilled in the
computer art. Appropriate software coding can readily be prepared
by skilled programmers based on the teachings of the present
disclosure, as will be apparent to those skilled in the software
art. The invention may also be implemented by the preparation of
integrated circuits or by interconnecting an appropriate network of
conventional component circuits, as will be readily apparent to
those skilled in the art.
[0110] Various embodiments include a computer program product which
is a storage medium (media) having instructions stored thereon/in
which can be used to program a general purpose or specialized
computing processor/device to perform any of the features presented
herein. The storage medium can include, but is not limited to, one
or more of the following: any type of physical media including
floppy disks, optical discs, DVDs, CD-ROMs, microdrives,
magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs,
flash memory devices, magnetic or optical cards, nanosystems
(including molecular memory ICs); and any type of media or device
suitable for storing instructions and/or data. Various embodiments
include a computer program product that can be transmitted over one
or more public and/or private networks wherein the transmission
includes instructions which can be used to program a computing
device to perform any of the features presented herein.
[0111] Stored one or more of the computer readable medium (media),
the present disclosure includes software for controlling both the
hardware of the general purpose/specialized computer or
microprocessor, and for enabling the computer or microprocessor to
interact with a human user or other mechanism utilizing the results
of the present invention. Such software may include, but is not
limited to, device drivers, operating systems, execution
environments/containers, and applications.
Summary
[0112] A brain-based device (BBD), including NOMAD 10 controlled by
a simulated nervous system 12 has been discussed, which bound the
visual attributes of distinct stimuli. Binding in the brain-based
device BBD occurred as a result of multilevel interactions
involving a reentrant neuroanatomy (FIG. 2, Table 1 and Table 2),
the dynamic synchronization of neuronal groups, and the
correlations generated by synaptic plasticity and autonomous
behavior of NOMAD 10 moving in its environment. Specifically,
during approaches to visual objects the formation of synchronously
active neuronal circuits occurred for each object in the visual
field of NOMAD 10. These circuits, which were enabled by reentrant
connections within and among neural areas V1, V2, etc., gave rise
to motor area activity which in turn evoked discriminatory behavior
of NOMAD 10. This provides insight into the complex, dynamic
interactions between brain, body, and behavior that underlie
effective visual object recognition.
[0113] The brain-based device BBD of the present invention has
innately specified behavior (i.e. tracking towards auditory or
visual stimuli) and innately specified value or salience for
certain environmental signals (e.g. positive value of sound). The
BBD learned autonomously to associate the value of the sound with
the attributes of the visual stimulus closest to the sound source,
and, it successfully oriented towards the target object based on
visual attributes alone (see FIG. 7A).
[0114] The physical embodiment of the brain-based device was
important for incorporating many of the challenging aspects of this
object discrimination task, such as variations in the position,
scale and luminosity of visual images, sound reflections, and
slippages during movement. Reliance on elaborate computer
simulations risks introducing a priori biases in the form of
implicit instructions governing interactions between an agent and
its environment. By the use of a real-world environment, however,
not only is the risk of introducing such biases avoided, but also
the need for the construction of a highly complex simulated
environment is eliminated.
[0115] The simulated nervous system 12 of the present invention
contains cortical areas analogous to the ventral occipito-temporal
stream of the visual system (areas V2, V4, and IT), the motor
system (area C), as well as reward or value systems (area S)
analogous to diffuse ascending neuromodulatory systems. None of
these specialized areas, however, nor preferential directions of
information flow (e.g. "top-down" or "bottom-up"), are by
themselves sufficient for binding the features of visual objects.
Rather, visual binding in the brain-based device BBD is achieved
through the interaction of local processes (i.e. activity in each
simulated neural area), and global processes (i.e. emergent
functional circuits characterized by synchronous activity
distributed throughout the simulated nervous system 12). Reentrant
connections among distributed neural areas V1, V2, etc. were found
to be essential for the formation of these circuits (see FIGS. 9,
10, and 12) and for successful performance in a task requiring
discrimination between multiple objects with shared features (see
FIG. 7). The brain-based device BBD of the present invention
achieved reliable discriminations in the visual field, which
resulted from self-generated or autonomous movement in a rich
real-world environment (see FIG. 11).
[0116] The state of each neuronal unit in the simulated nervous
system 12 has been described by both a firing rate variable and a
phase variable, where post-synaptic phase tends to be correlated
with the phase of the most strongly active pre-synaptic inputs.
This modeling strategy provided the temporal precision needed to
represent neural synchrony, without incurring the computational
costs associated with modeling of the spiking activity of
individual neurons. While representation of precise spike timing is
necessary for modeling certain neuronal interactions, the disclosed
model suggests that for the purposes of illustrating the mechanism
for visual binding, such detail is not required. It is also
important to emphasize that phase in the described model is not
intended as a reflection of possible underlying oscillatory
activity, specifically, it should not be taken to imply that
regular brain oscillations at specific frequencies are an essential
component of the neural mechanisms of binding.
[0117] Although local regions in the simulated nervous system 12
had segregated functions based on their input and connectivity,
object recognition and object discriminative behavior was an
emergent property of the whole system, not of any individual area.
The neural responses of the brain-based device BBD during an
orienting movement toward a target showed this global property in
terms of synchronized activity among a dynamic set of neuronal
units in different neural areas (see FIGS. 8 and 9A). The
simultaneous viewing of two objects clearly evoked two distinct
sets of circuits that were distributed throughout the simulated
nervous system 12 and distinguished by differences in the relative
timing of their activity. When the reentrant connections between
neural areas V1, V2, etc. were removed via simulated lesions,
coherent interactions among these neural areas were disrupted (see
FIGS. 9B, 10B, and 10C) resulting in failures in both object
perceptual categorization and object discriminative behavior (see
FIG. 7B).
[0118] Both experience and value shape the global properties of the
simulated nervous system 12. This is clearly shown in FIGS. 12A and
12B where, during early training, area S showed no activity, and
area C showed no bias toward the target object until the onset of
the auditory cue, i.e. value. Late in the training, area S became
active well before the auditory cue onset as a result of the
value-dependent plastic connections from area IT to area S, i.e.
value of visual stimuli. Activity in area S therefore became
predictive of the unconditioned stimuli (i.e. the auditory tone).
Value-dependent plastic connections from area IT to area C and
excitatory connections from area S to area C ensured that this
shift in the timing of value-related activity resulted in a bias in
the activity of area C which favored movement toward the target in
preference to the distracter. This emphasizes the role of value
systems in modifying the efficacy of distributed neural connections
to assure adaptive behavior. Successful performance in the object
discrimination task amongst objects in a visual field required the
complementary action of neural synchrony and experience-dependent
changes in neuronal firing rates (see Table 3). Neuronal synchrony,
which was indicated by groups of neuronal units sharing a similar
phase, was necessary for the formation of multiple global circuits
corresponding to each object in view. At the same time, the
activity of the neuronal units within these circuits influenced
activity levels in areas V4, IT, and C causing NOMAD 10 to favor
the target object over distracters. These observations suggest that
mean firing rate "codes" and synchrony-based "codes" need not be
considered as mutually exclusive explanations of neuronal
function.
[0119] A prediction of the described model, in which neuronal units
represent the activity of small groups of neurons, is that neural
synchrony at the group level, rather than zero phase lag among
individual neurons, may be sufficient for sensory binding. Although
some single-unit recording studies have shown that neurons
activated by attended stimuli are more synchronized than neurons
activated by unattended stimuli, synchronous activity among single
units has been difficult to detect in tasks requiring binding.
Also, micro-electrode recordings from primate prefrontal cortex
have shown higher levels of correlated firing among local,
inhibitory neurons than among excitatory, long-range pyramidal
neurons. On the other hand, neuromagnetic recordings of human
subjects during binocular rivalry have shown an increase in the
intra- and inter-hemispheric coherence of signals associated with a
perceptually dominant stimulus, as compared to a stimulus which is
not consciously perceived. However, neuromagnetic signals do not
reflect reentrant relations between single neurons; rather, they
represent averages across large neuronal populations. This is
therefore consistent with the model of the present invention
described above in suggesting that synchrony can operate at a
neuronal group level as well as at the single neuron level.
[0120] Higher brain function depends on the cooperative activity of
the entire nervous system, reflecting its morphology, its dynamics,
and its interactions with the body and the environment. In accord
with theoretical views emphasizing the importance of binding
through synchrony the brain-based device BBD of the present
invention shows that visual binding and object discrimination can
arise as a result of the constraints reentry and behavior impose on
interactions between local processes (activity in particular neural
areas) and global processes (synchronously active and broadly
distributed neural circuits). This interaction between these
processes was essential, and neither specialized areas nor
deterministic preferential directions of information flow were
sufficient alone to achieve visual binding.
[0121] The foregoing description of the preferred embodiments of
the present invention has been provided for purposes of
illustration and description. It is not intended to be exhaustive
or to limit the invention to the precise forms disclosed. Many
modifications and variations will be apparent to the practitioner
skilled in the art. Embodiments were chosen and described in order
to best explain the principles of the invention and its practical
application, thereby enabling others skilled in the art to
understand the invention, the various embodiments and with various
modifications that are suited to the particular use contemplated.
It is intended that the scope of the invention be defined by the
following claims and their equivalents.
* * * * *