The Biology of the Mind
Humans are animals. If we want to understand ourselves, we should understand
them.
Active embodiment and "scenes", or object-and-action schemas are the
keys to this.
Plan for this Lecture
arguments for the identity of human and animal minds
what evolutionary explanations tell about both
animal intentionality is the world of actions
active embodiment is the source of mental content
coming to a general picture of cognition
1. Arguments for the Identity of Human and Animal Mind
1.1. The Biological Function of Intelligence and its Characterization
In general:
Sensorimotor adaptation - not necessarily only in
the sense of optimality...
More closely:
Successful coordination of biological actions.
"Coordination" is a tricky word here - it anticipates our later conclusion
that much of high-level intelligence is based on low-level one.
For instance, having preexsiting primitive actions is important; the
rest is just to put them together, to "co-ordinate" them, in a minimalist
sense.
In detail: 1.2. and 1.3.
1.2. Human Behavior Serves Biological Function
Explanatory illustration: an "anecdote" taken from a famous book.
Emmanuele Le Roy Ladurie: "Montaillou" (New York: Vintage, 1979)
Peasants in 13. century France spent much of their time just scratching
- as baboons do ("grooming").
(in fact not anecdote a revealing story from microhistory and its coupling
to a known ethological fact about primates)
About "Montaillou"
"This
book tells the story of the 14th century villagers and peasants of Montaillou.
It chronicles their everyday lives,
its importance is far greater than the story of the Cathar heresy, it is
the story of the kind of people that are usually
forgotten
to history."
"love and marriage, gestures and emotions, conversations and gossip, clans
and factions, crime and violence, concepts of time
and
space, attitudes to the past, animals, magic and folklore, death and beliefs
about the other world."
1.3. What the Mind of an Organism is for
"Elephants Don't Play Chess"
(R. Brooks). Nor did humans when they acquired minds.
Sensory input occurs in the biological context:
in meaningful situations of complex
nature
not raw input but based on co-existence
and mutual interaction
Remember the notion of language game - this is also a "game"
Perhaps the least unit is a meaningful complex of actions
We will resolve the holistic "situation" of this "game" into composites:
origin of mechanisms, origin of logic, origin of narratives.
1.4. Same Function - Same Structure
Animal and human minds serve the same biological functions and operate among
the same essential circumstances.
Is it the same mind? We have to look into evolution to see what kind of
mind it is (and this will supply more details
on why to believe in the identity).
2. What Evolutionary Explanations Tell About the Mind
2.1. The Nature of Evolutionary Explanation
Search for the distal causes of behavior control
Note Mayr's distinction between proximal (= immediate, close) and distal
(= past, remote) causes in evolution
Example for distal cause is adaptation.
A well-known application of this principle is sociobiology and evolutionary
psychology
They have an exaggerated focus on adaptation.
(2.2. The Positive Message from all Evolutionary Explanations
Psychological properies are biological properties, which have a history.
History can explain the biological basis and function of cognition.
....This picture of the mind is usually completely ignored (not even denied!)
in mainstream knowledge & mind theory.)
2.3. What is Evolution if not Adaptation
Adaptation is an extremely (!) important but NOT a universal evolutionary
feature.
Three universal evolutionary features: unity, embodiment, entrenchment.
Unity, def:
All animals are similar in all traits, the closer
relatives the better. A consequence of evolutionary continuity.
Embodiment, def.: The input of the
mind is what the body produces when interacting with the environment
Entrenchment, def.: Old systems are
"buried" under new ones, and impossible to change.
2.4. Unity (works both ways)
Species exist in philogenetic continuity.
Therefore,
(1) Most human traits must exist in animals.
(2) Most animal traits must exist in man.
Both are big conceptual steps forward - (1) the lifting of animals to humans
(2) the lowering of man to animals.
There is a temptation to believe that typical human cognitive abilities
are very high-level and specific to us,
and to believe at the same time, that animal abilities are low level ones.
Unity shows that it cannot be that way.
This is a very simple point
which is overlooked so often.
Of course this point
could be elaborated in detail.
The argument from unity
assumes gradualism of all traits, for instance.
In a more elaborate treatment
we should consider philogenetic mechanisms etc., but that does not change
much.
2.5. Entrenchment
I.e. The burying of old systems
The metaphor: layers of sand and soil and debris.... as in archeology
(W. Wimsatt)
Schank, J. C., and W. C. Wimsatt, 1988. Generative entrenchment
and evolution.
In: A. Fine and P.
K. Machamer, eds., PSA-1986, vol. 2. East Lansing: The Philosophy of Science
Association, pp. 33-60.
What is buried away cannot change later, bacause it is no more accessible
for change.
Less metaphorically: all mutationsare devastating because of avalanche-like
cumulative effects.
If we find an old system in humans, it must be exactly (!) the same in
even distant animals.
(Of course here we use the fact that historical distance implies present-day
distance).
Much of the human cognitive systems is old.
Therefore, the human system must be identical with the old animal systems
which we
find in distant species today.
There is an
important side issue here. The concept of entrenchment is a developmental
one, and it does not directly apply to immediate gene products.
So the well-known
examples for "molecular clocks in evolution" (changes
at a steady rate) do not contradict the argument.
How do we know that human cognitive systems are old?
There are various justifications of
this notion. Such as:
Examples from complex physiological
adaptations. E.g. sneezing.
Sneezing is an old system found in many mammals.
It always produces the same typical
pattern: eyes shut, hight muscle tone, etc.
From entrenchment we understand
that one feature cannot be changed later.
Therefore, speezing must also produce
the same chilly feeling in dogs and cats etc.
This is not the best example for cognition (sneezing is not very smart
eh?) but it is vey compellig.
A better (but less dramatic) example
is social facilitation in dogs.
Dogs can spontaneously learn pointing
and eye-following.
This has a twis, as dogs are human
products selected for human-like features.
Why is the story relevant, then?
(1) But they remained very distant
relatives to humans.
If they - with some evolutrionary
modifications - can learn these, then the same basic system
of interpretation is there in the
mind since the common ancestors of dog and man.
(2) Knock-down argument: but dogs
are still animals, arent't they?
So probably other animals are capable
of the same, with a little breeding help from man.
Another (and, again, more dramatic) example will be discussed with in
relationship to "agency".
2.6. Evolutionary Embodiment
The basis:
Genotype - phenotype (replicator - interactor). Every evolutionary interaction
is based on the phenotpye.
The phenotype defines boundary conditions of interaction (in the air, in
the water, on the soil etc).
There is no "environment" in the abstract sense - note that historically
the word "Umwelt" (von Uexkull) reflects this.
The crosspoint of environment and mind (or gene) is the actually existing
body.
Mental content cannot be (is not) a "representation" of environment as
something completely idependent
mental content consists of utilization recipes for the
available joint features of the enviroment and the body.
It is the body where meaning resides, if seen on the evolutionary context.
Language and other concepts "plug in" into ready-made meanings available
from philogenetically earlier times.
Such meaning must be non-verbal, non-conceptual, at least backwards from
some point of evolutionary history
(Much research
is done these days on embodiment and "non-conceptual content" etc. so part
of this is fairly
standard material - up to the point where I will speak of active embodiment)
NCC:
Definitions vary, but all writers agree on this: a mental state's
content is non-conceptual if an organism can be in
that state without having to possess the concepts used in canonically characterizing
that state's content.
"What is Non-Conceptual Content?" http://www.cogs.susx.ac.uk/users/ronc/whatis-ncc.html
A bibliography of non-conceptual content http://www.cogs.susx.ac.uk/users/ronc/ncc-bibliography.htm
More about embodiment comes below.
3. Inside the Animal Mind
the strongest pillars are these: inborn intentionality; intentionality
is coupled to agency; animal agency detection
3.1. Intentionality
Intentionality, definition: beliefs, desires etc and their attribution
(one of the evergreens
of cognitive science, http://plato.stanford.edu/entries/consciousness-intentionality/
http://www.u.arizona.edu/~chalmers/biblio.html;
in particular http://www.u.arizona.edu/~chalmers/biblio/2.html)
There is a vast amount of evidence about the existence of animal intentionality,
with details unclear.
consciousness
tests http://plato.stanford.edu/entries/consciousness-animal/;
http://www.u.arizona.edu/~chalmers/biblio/6.html#6.4c
imitation (social
facilitation), viewpoint taking (e.g. triangulation), animal lies (misguiding
behavior)
naming (categorization
plus "rigid designator") and more.....
- adaptation,
teleology, D.C. Dennett: The Intentional Stance (1987), Cambridge
MA, MIT Press; http://cognet.mit.edu/MITECS/Entry/dennett
- a comprehesive
treatment of the full range: Allan and Beckoff: Species of Mind
full text
Important, however:
Intentionality's roots are inborn in humans (Meltzoff; Gergely and Watson)
Meltzoff early
imitation experiments showed that already newborns have "social" mental competence
description of the Meltzoff experiments.
Gergely and Watson
show that 3 months old infants have firm expectations about actions
description of the Gergely and Watson experiments
The apparent basis of early intentionality of the latter type is the use
of built-in contingency detectors (Gergely):
stimuli with high
contingency (high reliablity) are preferred in age 0-3 month;
low contingency
(random) is ignored both then and later;
interest switches
to medium contingency at 3 months
medium contingency is typical for "agents" (actors) with voluntary actions
(autonomy)
To sum up: intentionality reduces to agency and has a clear evolutionary
origin ("inborn" means this).
A remark. In developmental psychology a conceptually more refined picture
is used.
The relationship between (early) intentionality and agency
is a matter of ongoing debate.
The typical view is that intentionality is part of the
"self" concept and agency also, but the
two are not directly related (there is an indirect
relation throught the self).
3.2. Agency, definition
Agents: actors, pro-active entities, have the ability to initiate action;
agents are movers (cf. what moves what).
3.3. Agency detection by animals
This is a hard problem, since movers and the moved parts are highly correlated
Perhaps contingency is a cue, again.
The theoretical explanation for agency
detection is not known yet.
The empirical facts, however, are well-studied - in dogs and cats, for
instance.
Dogs are able to launch tennis ball by their nose upon observing
humans doing it by hand
(pictures pending)
Cats open doors by jumping up and pulling the door-handle, upon observing
humans turning it
(picture pending)
3.4. Objects and Actions in Animal Life
It appears that the basic elements of the animal mental world are objects
and actions.
Is this perhaps a self-evident truth? What else could they be?
In actuality, it is far from being a safe and/or simple statement, but
it can be risked.
Objects - it is generally assummed / accepted that animals
(i.e. higher animals) partition their environment into objects
(for instance,
every study on animal intentionality and agency uses this as an unproblematic
fact).
Actions - we just discussed that actions are performed
by agents, and agents are not easy to define or identify.
In fact
both finding objects and findig the agents in the environment are difficult
problems for robotics,
even if it is assumed
that these are important.
Objects are special integrative wholes, and actions similarly so.
Instead of dealing with their sensory and neural backgrounds, we will focus
on these questions:
what properties
of the environment make object - and - action thinking possible?
how does object
- and - action thinking work?
what consequences
are there for the structure of the animal and human mind?
3.5. The 'Scene' Theory: Dynamic Situations
In particular, we can ask about the relationship between festures of the
structure of environment and features of mind.
The remarks here will be rudimentary and hypothetical.
I put forward a scene theory which combines many elements that are
already at hand.
Adult animals (and humans) organize their perceptual fields into scenes
(plots, scripts, situations, sessions...)
Scene = an interactive bevavior complex, which involves a goal (or a main
motif), some actor(s) and actions, objects, and
spatial as well
as temporal elements.
A scene is the least contiguous unit for mental activity (---- > this
will be captured in the notion of mental models)
Evidence for the "scene" theory
Tolman's expectation
theory
(animals anticipate sensory
input as part of a larger unit)
Thorndike's ecological
selection theory (animals
select relevant signals in the environment on an ecological basis; not anything
is a signal)
The "verb" theory
of dog behavior
(dogs - and other animals - take words not
as names but as 'verbs' relative to behavior complexes)
4. Embodiment and Active Embodiment
Embodiment is one of the big stories of the nineties:
How the mind uses the body, and how the body is reflected in the mind.
Wheeler, M. (1997): Cognition's Coming Home: The Reunion of Life and
Mind, in:
P. Husbands és I. Harvey szerk.: Fourth European Conference on Artificial
Life,
MIT Press, Cambridge, MA.
ftp://ftp.cogs.susx.ac.uk/pub/ecal97/online/F035.ps.gz
Sometimes there is a spiritual or existentialist turn here:
http://www.bodywisdom.org/pages/embod.html
4.1. The Robotics Version
Intelligence without representation - simple agents perform directional
operations using their body
No description, no "mind" (in the classical sense), only action control
("Cambrian intelligence").
http://www.ai.mit.edu/people/brooks/publications.shtml
Brooks, R. A., (1986), Achieving artificial intelligence through building
robots, MIT A. I. Memo No. 899.
Brooks, R. A., (1991a), New approaches to robotics, Science, 253 , 1227–1232.
Brooks, R. A., (1991b), Intelligence without reason. Proceedings of IJCAI–91,
569–595,
Brooks, R. A. és Stein, L. A. (1993), „Building brains for bodies”,
MIT A. I. Memo No. 1439.
Brooks, R.A., Breazeal, C., Irie, R., Kemp, C.C., Marjanovic, M., Scassellati,
B.,Williamson, M. (1998): Alternate Essences of Intelligence, Proc. AAAI-98
Brooks, R.A, Breazeal, C., Marjanovic, M., Scassellati, M., Williamson,
M. (1998): The Cog Project: Building a Humanoid Robot, in: (C. Nehaniv, C.
szerk.)
Computation for Metaphors, Analogy and Agents, Vol.
1562, Springer Lecture Notes in Artificial Intelligence, Springer-Verlag,
New York.
4.2. Beyond Robotics: Coordination Revisited
Brooks et al: embodiment reduces to "action selection".
From the persepctive of cognitive ethology of higher animals, this is not
cognition itself but the 'raw material' for that.
Cognition is coordination, but it is not jut reactive.
4.3. The Meaning of Embodiment for Cognition
A bodily basis for the mental.
Example: The Theory of Metaphor (Lakoff and Johnson)
Lakoff, G. (1987): Women, Fire, and Dangerous Things, Chicago UP,
Chicago, IL.
Lakoff, G. és Johnson, M. (1980): Metaphors We Live By, Chicago
UP, Chicago, IL.
Johnson, M. (1987): The Body in the Mind, Chicago UP, Chicago, IL.
Annotated Bibliography
of Metaphor and Cognitive science
Center for
the Cognitive Science of Metaphor, Online
The basic idea: a small number of fundamental image schemas (Gestalt) bear
all meaning in language ----> all meaning in the mind
Container
|
Balance
|
Full-Empty
|
Iteration
|
Compulsion
|
Blockage
|
Counterforce
|
Process
|
Surface
|
Restraint Removal
|
Enablement
|
Attraction
|
Matching
|
Part-Whole
|
Mass-Count
|
Path
|
Link
|
Collection
|
Contact
|
Center-Periphery
|
Cycle
|
Splitting
|
Merging
|
Object
|
Scale
|
Table: The partial list of image schemata from
Johnson (1987)
Metaphor = figurative use of basic schemas (e.g. into, in, prevent, etc
etc)
George Lakoff: The Contemporary Theory of Metaphor, full text here
4.4. Active Embodiment
The situated/embodied picture is based on sensorimotor notions, yet sometimes
the resulting image of the mind is passive.
For instance, a focus on experience puts the mind in the position of a static
observer; things happen to the owner of this mind.
Most image schemas of Johnson are just descriptive, even if (as in the case
of force) this is not obvious at the first glance.
Perhaps garden-path and a few others are exceptions. But even these are
more contemplative trhan not.
The fundamental schemas of action, such as an if.... then schema
are completely missing.
Work in cognitive developmental psychology on the origin of embodied concepts
somewhat changes this situation.
Thelen et al characterize "force embodiment" as a concept resulting from
cycles of repeated proactivity and related experience.
The "scene" theory extends this into a general framework of active embodiment,
where perception-action cycles organized
into meaningful units play the fundamental role.
5. Coming to a General Picture
5.1. The Strategy from Here
Knowledge of Animals
Thinking without words; a model of the animal mind
A pre-wired world, consisting objects and organisms, as source of animal
knowledge.
Body skills and their use in cognition.
The Everyday World
No knowledge without prior knowledge
Everyday realism, folk ontology, folk psychology.
The reliable world as human and animal legacy
Mechanisms as simplified causal schemes.
Rationality in Action
Rationality's basic form: goal-consistent action plan.
Abstract knowledge as based on action: the active mind.
etc/
5.2. Relationship with Human Knowledge; Backflash to Lecture Two.
Wittgenstein is so special not because no one can argue against it.
(In fact many philosophers, including Kripke, Putnam, and the new analitics
do that.)\
That is just part of the usual game of philosophy.
But Wittgenstein's significance is different: it's the only philosophical
picture of the mind
that is compatible with philogenetic continuity: unity,
embodiment and entrenchment.
Remark:
Here it will be important to ask how certain our knowledge about evolution
is.
That could be the topic of another lecture.
The answer, in short, is this: very certain - in fact as certain as
you can get.
5.3. Summary: the "Discovery" of Animals
Cognitive science is now in the process of (re)discovering animals.
But: Native Americans don't like to hear that they were "discovered" by
Columbus... and if animals had abstract
concepts, they would probably raise the same objection.
Animals were always in front of our eyes.
More than just that, humans have always lived together with animals
- the dog is about as old as man.
Breeders and keepers have treated animals in everyday interaction and
accummulated evidence on animal minds.
(cf. The meaning of history: compare selected writers to what large numbers
of people think/know in a period.
Microsociology/microhistory - another revolution in the making...)
Motto:
Darwin grounded evolution theory on what every breeder knew but no philosopher
did.
Cognitive science can ground the theory of mind in the same way.