Transcript Mechanism for Consciousness

A Mechanism for Learning, Attention
Switching, and Consciousness
Janusz Starzyk
School of Electrical Engineering and Computer Science,
Ohio University, USA
http://people.ohio.edu/starzykj
October 20, 2010.
Outline









Attention
Biological perspective
Emergence of consciousness
Functional requirements
Computational model of consciousness
Attention switching
Mental saccades
Implications
Summary
Photo: https://www.adbusters.org/magazine/87/philosophy-zero-point.html
Big Questions
How a single thought emerges
in your brain?
 What motivates you to learn or
do anything?
 How can you switch your attention from one activity to
another?
 What is necessary for cognition, intelligence, and
consciousness?
 These are but few questions important to philosophers,
cognitive neuroscientists, psychologists, artificial intelligence
researchers, etc.

Photo: http://tsvetankapetrova.wordpress.com/2009/06/30/5-setbacks-that-stop-you-big-time/
Big Questions
Can computational models be
provided that demonstrate some of
these phenomena?
 Can we make a practical use of
them in autonomous machines
working in real time in natural
environments?
 This talk will address some of
these questions.

http://www.geneang.com/Presence_Healing,_LLC/Neuroscience_of_Consciousness.html
Attention
 The term attention is used when there is a clear voluntary act.
 We ask people to pay attention and they can chose to do so or not.
 Voluntary attention is involved in preparing and applying goal directed
selection for stimuli and responses.
5
Attention
 Attention selects information for cognitive process
 Selection is driven by perceptions, emotions, motivations and is under
executive control.
 Without flexible, voluntary attention, we would not be able to change
behavior or deal with unexpected emergencies or opportunities.
 Without stimulus-driven attention we would not be able to respond
quickly to significant external events.
 Thus we need both voluntary and automatic attention.
6
Brain basis of attention
 William James wrote
that attention helps to:





Perceive
Conceive
Distinguish
Remember
Shorten reaction time
 Attention to a location
improves the accuracy
and speed of detecting
target at this location.
 Attention can be based on internal goals (finding a friend in the
crowd) or external environment (alarm, bright colors)
7
Brain basis of attention
 Maintaining attention against distraction requires a significant
effort;
 E.g. trying to study when your roommate plays a loud music
 Thus mental effort comes from struggle between voluntary (goal
driven) and automatic attention.
8
Brain basis of consciousness
 Conscious cognition is close to attention, but not the same.
 You can tell people – please pay attention but not - please be conscious.
 You may be aware (conscious) of reading this text but you may be not
aware of the touch of your chair, gravitational forces, background
conversation, your feelings for a friend, or your major life goals.
 Consciousness is not just a passive experience of sensory inputs, but an
active involvement and perception.
 “Self "-related phenomena such as preference, self-recognition, reflection,
and planning are central to an understanding of consciousness.
9
Consciousness

Differences between conscious and
unconscious phenomena
Conscious
 1. Explicit cognition
 2. Immediate memory
 3. Novel, informative, and
significant events
 4. Attended information
 5. Focal contents
 6. Declarative memory
(facts, etc.)
 7. Effortful tasks
 8. Remembering (recall)
 9. Available memories
Unconscious
Implicit cognition
Longer term memory
Routine, predictable,
and nonsignificant events
Unattended information
Fringe contents (e.g., familiarity)
Procedural memory
(skills, etc.)
Spontaneous/automatic tasks
Knowing (recognition)
Unavailable memories
10
Consciousness

Differences between conscious and
unconscious phenomena
Conscious
 10. Strategic control
 11. Grammatical strings
 12. Rehearsed items in
Working Memory
 13. Wakefulness and
dreams (cortical arousal)
 14. Explicit inferences
 15. Episodic memory
(autobiographical)
 16. Intentional learning
 17. Normal vision
Unconscious
Automatic control
Implicit underlying grammars
Unrehearsed items in
Working Memory
Deep sleep, coma, sedation
(cortical slow waves)
Automatic inferences
Semantic memory
(conceptual knowledge)
Incidental learning
Blindsight (cortical blindness)
11
Evolution and consciousness
– appearance and evolution of consciousness
Evolutionary traits
Analogous
feasibility in
machines
Human
 Fully developed cross-modal representation
Impossible at
Beings
 Sensory capabilities: auditory, taste, touch, vision, etc.
present
Living Being
 Pre-frontal cortex: planning, thought, motivation
Hedgehog
(earliest
mammals)
 Cross-modal representation
 Sensory capabilities: auditory, touch, vision (less
Impossible at
developed), etc.
present
 Small frontal cortex
 Primitive cross-modal representation
Birds
 Sensory capabilities: auditory, touch, vision, olfactory.
 Primitive associative memory
Associative
memories
Photos: http://images.google.com/
Evolution and consciousness
– absence of consciousness
Living Being
Reptiles*
Hagfish (early
vertebrate)
Lower level
animals
(hydra, sponge, etc.)
Evolutionary traits
 Olfactory system
 Primitive vision
 Primitive olfactory system
 Primitive nervous system
 Sensory motor units
 Point to point nervous system
Analogous feasibility
in machines
Computer vision
(emerging)
Artificial neural
networks
Mechanical or
electronic control
systems
* inconclusive\consciousness in transition
Photos: http://images.google.com/
Emergence of Consciousness
Week
Human Fetus brain development
6
Cortical cells come at the correct position
20
Cortical region is insulated with myelin sheath
25
Development of local connections between neurons
30
Fetus’ brain generates electrical wave patterns
Photos: http://daymix.com/Fetus-Brain-Development/
http://www.humanillnesses.com/Behavioral-Health-A-Br/The-Brain-and-Nervous-System.html?Comments[do]=mod&Comments[id]=1
Emergence of Consciousness

Brain is self-organizing and sparse
Human Brain
at Birth
6 Years Old
14 Years Old
Rethinking the Brain, Families and Work Institute, Rima Shore, 1997.
Synaptic Density over the Lifespan
Conclusion : Consciousness emerges gradually
Thompson, R. A., & Nelson, C. A. (2001). Developmental science and the media: Early brain development. American Psychologist, 56(1), 5-15.
Frontal lobe functions
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
Planning, setting goals and initiating actions
Monitoring outcomes and adapting to errors
Mental effort in pursuing difficult goals
Having motivations
Initiating speech and visual imagery
Recognizing other’s people’s goals
Engaging in social cooperation and competition
Feeling and regulating emotions
Storing and updating working memory
Active thinking
Enabling conscious experiences
Sustained attention in the face of distraction
Switching attention
Decision making and changing strategies
Planning and sequencing actions
Unifying the syntax and meaning of language
Resolving competition between plans
csiwebcomics.com
17
Description of Consciousness

Nobody has a slightest idea of how anything material can
be conscious
– Jerry Alan Fodor prof. of philosophy and cognitive science at Rutgers

The quality or state of being aware especially of something
within oneself
- Merriam Webster Dictionary

…our subjective experience or conscious state involving
awareness, attention, and self reference
– prof. Jeanette Norden neuroscientist in Vanderbilt .
 Consciousness is a dynamic process and it changes with
development of brain. Further, at macro-level there is no
consciousness centre and at micro-level there are no
committed neurons or genes dedicated to consciousness
– prof. Susan Greenfield neuroscientist director of Royal Institution GB
Conscious System Requirements
1.
2.
3.
4.
5.
6.
7.
8.
9.
Conscious system is aware of past and present and is capable
of critical analysis;
is aware of the environment in which it resides;
has a perception of its internal states
is capable to predict and explain current and past events;
is capable of autonomous construction of future actions;
can utilize past actions in the formulation of future plans:
is able to locate itself in its relationship to other entities;
can generate an internal representation of itself and its
environment
is capable of autonomously and selectively directing its
attention to address current important situations.
Neural model for consciousness
 A neural net architecture for
attention and visual
consciousness.

Visual information flows
from V1 to areas V2-V4,
and finally IT where
objects are detected.
 Each area has its inhibitory
neurons to sharpen
differences at that level.


Posterior parietal neurons
(PP) bias visual neurons
that detect the object in
that spatial location.
Prefrontal neurons in area
46 are involved in
voluntary attentional
selection.
Attention and conscious flows.
20
Proposed approach to
machine consciousness
 Define consciousness in
functional terms
 Identify minimum functional
requirements for consciousness
 Identify functional blocks, their
roles, their inter-relationships
 Propose a computational model
of a conscious machine
Photo: http://www.theglobalintelligencer.com/aug2007/fringe
Definition of Machine Consciousness
Consciousness is attention driven cognitive perception
motivations, thoughts, plans and action monitoring.
A machine is conscious IFF besides ability to perceive, act,
learn and remember, it has a central executive mechanism
that controls all the processes (conscious or subconscious)
of the machine;
Photo: www.spectrum.ieee.org/.../biorobot11f-thumb.jpg
Consciousness:
 Consciousness requires
– Intelligence (ability)
– Awareness (state)
 Not necessary alive
 How to model
consciousness?
Consciousness:
functional requirements
 Intelligence
 Central executive
 Attention and attention switching
 Mental saccades
 Cognitive perception
 Cognitive action control
Photo: http://eduspaces.net/csessums/weblog/11712.html
Computational Model of Machine Consciousness
Episodic
Memory &
Learning
Planning and
thinking
Central
Executive
Queuing and
organization
of episodes
Attention
switching
Motivation and
goal processor
Action
monitoring
Episodic
memory
Semantic
memory
Emotions, rewards, and
sub-cortical processing
Motor
skills
Sensory
processors
Motor
processors
Sensory-motor
Inspiration: human brain
Data encoders/ decoders
Data encoders/ decoders
Sensory
units
Motor
units
Photo (brain): http://www.scholarpedia.org/article/Neuronal_correlates_of_consciousness
Sensory and Motor Hierarchies

Sensory and motor
systems appear to be
arranged in hierarchies
with information
flowing between each
level of the sensory
and motor hierarchies.
26
Sensory- Motor Block
Semantic
memory
Emotions, rewards, and
sub-cortical processing
Sensory
processors
Motor skills
Motor
processors
Sensory-motor
Data encoders/ decoders
Data encoders/ decoders
Sensory units
Motor units
 sensory processors integrated with semantic memory
 motor processors integrated with motor skills
 sub-cortical processors integrated with emotions and rewards
Central Executive
 Platform for the emergence, control, and manifestation of
consciousness
 Controls its conscious and subconscious processes
 Is driven by



attention switching
learning mechanism
creation and selection of
motivations and goals
ahsmail.uwaterloo.ca/kin356/cexec/cexec.htm
Central Executive
Planning and
thinking
Motivation and
goal processor
Attention
switching
Central
Executive
Action
monitoring
 Tasks
o
o
o
o
o
o
o
o
cognitive perception
attention
attention switching
motivation
goal creation and selection
thoughts
planning
learning, etc.
Central Executive
Planning and
thinking
Attention
switching
Motivation and
goal processor
Central
Executive
Action
monitoring
 Interacts with other units for
o
o
o
performing its tasks
gathering data
giving directions to other units
 No clearly identified decision center
 Decisions are influenced by
o
o
competing signals representing motivations, pains, desires,
plans, and interrupt signals
• need not be cognitive or consciously realized
competition can be interrupted by attention switching signal
Attention Switching !!!
 Attention

is a selective process of cognitive
perception, action and other cognitive
experiences like
thoughts, action planning,
expectations, dreams
 Attention switching
 is needed to have a cognitive experience
 leads to sequences of cognitive
experiences
Comic: http://lonewolflibrarian.wordpress.com/2009/08/05/attention-and-distraction-what-are-you-paying-attention-to-08-05-09/
Attention Switching !!!
 Dynamic process resulting from competition between
• representations related to motivations
• sensory inputs
• internal thoughts including spurious signals (like noise).
blog.gigoo.org/.../
Attention Switching !!!
May be a result of :
•deliberate cognitive
experience (and thus
fully conscious signal)
• subconscious process
(stimulated by internal or
external signals)
Thus, while paying attention is a conscious experience,
switching attention does not have to be.
Mental Saccades
Memory traces in frontal cortex
 Selected part of
the image
resulting from an
eye saccade.
 Perceived input
activates object
recognition and
associated areas
of semantic and
episodic memory.
Frontal cortex
wife
friends
Episodic and
associative
memory network
his wife
friends
Input image
house
business
Spotlight
on John
business
John
dog
his house
his dog
saccade
 This in turn activates memory traces in the global workspace area
that will be used for mental searches (mental saccades).
Mental
saccade
Mental saccades in a conscious machine
No
Advancement
of a goal?
Attention spotlight
Yes
Loop 1
Mental saccades
Learning
Yes
Continue
search?
Changing motivation
No
Loop 3
Plan action?
Associative memory
Yes
Loop 2
No
No
Action?
Changing perception
Yes
Action control
Loop 4
Loop 5
Changing environment
Perceptual saccades
Computational Model: Summary
 Self-organizing mechanism of emerging motivations
and other signals competing for attention is fundamental
for conscious machines.
 A central executive controls conscious and
subconscious processes driven by its attention switching
mechanism.
 Attention switching is a dynamic process resulting from
competition between representations, sensory inputs and
internal thoughts
 Mental saccades of the working memory are
fundamental for cognitive thinking, attention switching,
planning, and action monitoring
Photo: http://www.prlog.org/10313829-homeless-man-earns-250000-after-viewing-prosperity-consciousness-video-subliminal-mindtraining.html
Computational Model: Implications
 Motivations for actions are physically distributed
o
competing signals are generated in various parts of machine’s
mind
 Before a winner is selected, machine does not interpret
the meaning of the competing signals
 Cognitive processing is predominantly sequential
o
winner of the internal competition is an instantaneous director
of the cognitive thought process, before it is replaced by
another winner
 Top down activation for perception, planning, internal
thought or motor functions
o
results in conscious experience
•
•
o
decision of what is observed and where is it
planning how to respond
a train of such experiences constitutes consciousness
Conclusions
1. Consciousness is computational
2. Intelligent machines can be conscious
Sounds like science fiction?


If you’re trying to look far
ahead, and what you see
seems like science fiction,
it might be wrong.
But if it doesn’t seem like
science fiction, it’s
definitely wrong.
From presentation by Feresight Institute
Questions ??
Photo: http://bajan.wordpress.com/2010/03/03/dont-blame-life-blame-the-way-how-you-live-it/
References











J. A. Fodor, "The big idea: can there be science of the mind," Times
Literary Supplement, pp. 5-7, July 1992.
J. Norden, Understanding the brain, Video lecture series.
M. Velmans, "Where experiences are: Dualist, physicalist, enactive and
reflexive accounts of phenomenal consciousness," Phenomenology
and the Cognitive Sciences, vol. 6, pp. 547-563, 2007
A. Sloman, "Developing concept of consciousness," Behavioral and
Brain Sciences, vol. 14 (4), pp. 694-695, Dec 1991.
W. H. Calvin and G. A. Ojemann, Conversation with Neil's brain: the
neural nature of thought and language: Addison-Wesley, 1994.
J. Hawkins and S. Blakeslee, On intelligence. New York: Henry Holt &
Company, LLC., 2004.
S. Greenfield, The private life of the brain. New York: John Wiley &
Sons, Inc., 2000.
Nisargadatta, I am that. Bombay: Chetana Publishing, 1973.
D. C. Dennett, Consciousness Explained, Penguin Press,1993.
D. M. Rosenthal, The nature of Mind, Oxford University Press, 1991.
B. J. Baars “A cognitive theory of consciousness,” Cambridge
University Press, 1998.
Photo: http://s121.photobucket.com/albums/o209/TiTekty/?action=view&current=hist_sci_image1.jpg
Embodied Intelligence
Definition
 Embodied Intelligence (EI) is a
mechanism that learns how to
minimize hostility of its environment
– Mechanism: biological, mechanical or virtual agent
with embodied sensors and actuators
– EI acts on environment and perceives its actions
– Environment hostility is persistent and stimulates EI to act
– Hostility: direct aggression, pain, scarce resources, etc
– EI learns so it must have associative self-organizing memory
– Knowledge is acquired by EI
Embodiment of a Mind

Embodiment is a part of the environment that EI controls to
interact with the rest of the environment
 It contains intelligence core
and sensory motor
interfaces under its control
 Necessary for development
of intelligence
 Not necessarily constant or
in the form of a physical
body
 Boundary transforms
modifying brain’s selfdetermination
Embodiment
Sensors
channel
Environment
Intelligence
core
Actuators
channel
Embodiment of a Mind




Brain learns own body’s dynamic
Self-awareness is a result of
identification with own embodiment
Embodiment can be extended by
using tools and machines
Successful operation is a function
of correct perception of
environment and own embodiment
Motivated Learning

Definition: Motivated learning (ML) is pain
based motivation, goal creation and learning in
embodied agent.






Machine creates abstract goals based on the
primitive pain signals.
It receives internal rewards for satisfying its goals
(both primitive and abstract).
ML applies to EI working in a hostile environment.
Various pains and external signals compete for attention.
Attention switching results from competition.
Cognitive perception is aided by winner of competition.
Reinforcement Learning


Single value function
Measurable rewards
 Can be optimized



Predictable
Objectives set by
designer
Maximizes the reward
Motivated Learning

 One for each goal


Learning effort increases
with complexity
Always active
Internal rewards
 Cannot be optimized



 Potentially unstable

Multiple value functions
Unpredictable
Sets its own objectives
Solves minimax problem
 Always stable


Learns better in complex
environment than RL
Acts when needed
http://www.bradfordvts.co.uk/images/goal.jpg