Re: [ontolog-forum] Body Parts and Early-Learned Verbs

[Gary Berg-Cross <gbergcross@xxxxxxxxx>]

The Piagetian idea that high-level, abstract cognitive competences can be grounded & bootstrapped from sensorimotor behavior is an ongoing frame for developmental hypotheses.  It may also be worth noting that it serves to frame work within several newly emerging & overlapping research field known variously as “embodied artificial intelligence,” “embodied cognition,” “developmental robotics,” or “epigenetic robotics. Developmental robotics, for example, is an interdisciplinary and empirical field that studies how autonomous, adaptation-oriented robots can learn to acquire behavior & knowledge on their own, strictly through their interactions with the surrounding environment. The field provides some interesting, empirical tests of various Piagetian inspired ideas, for example, of how verbal behavior develops out of sensorimotor schema. Early overviews of the field go back to: Lungarella et al., 2003; Pfeifer and Iida, 2004)

 Developmental robotics approaches typically start with some small number of built-in mechanisms perhaps a mix of reflexes, motivational drives, & some learning mechanisms such as abstraction). The autonomous robot-agent then is allowed to interact in a real physical environment. In the process the agent “discovers” relations between actions and sensations by this interacton. This learned relational knowledge is used in turn add to the built-in mechanisms and thus reach a first level of cognitive competences . You can see a good version of this idea in operation in (Blank et al’s., 2005)- Bringing up robot: Fundamental mechanisms for creating a self-motivated, self-organizing architecture. Cybernetics and Systems 2005 36 2 125–150.

 

But it is chicken and egg challenging since we have faint understanding of how innate mechanisms for abstraction, prediction, and self-motivation can be realized in such autonomous systems through interaction with the world including feedback from other intelligent agents – “yes, that is a dog.” 

Gary Berg-Cross, Ph.D.  

gbergcross@xxxxxxxxx     

http://ontolog.cim3.net/cgi-bin/wiki.pl?GaryBergCross

NSF INTEROP Project  

http://www.nsf.gov/awardsearch/showAward.do?AwardNumber=0955816

SOCoP Executive Secretary

Knowledge Strategies    

Potomac, MD

240-426-0770

On Tue, Sep 3, 2013 at 1:44 PM, Aldo Gangemi <gangemi@xxxxxxx> wrote:

On Sep 3, 2013, at 7:12:27 PM , John F Sowa <sowa@xxxxxxxxxxx> wrote:

> Joanne, Sandro, and Mike,
>
> JFS
>>> This study is one of many that show how the semantics of natural
>>> language is grounded in the neural mechanisms of perception and action.
>
> JL
>> Isn't that what the symbol grounding problem is all about?
>
> Yes. But I believe that the work on psycholinguistics and neuroscience
> is more relevant than philosophical debate about Chinese rooms.
>
> JL
>> This is a fascinating observation about relationships, and raises
>> interesting questions about whether and how well we can capture
>> context and culture in our current mapping schemes.
>
> That is the critical issue:  How is all the work on logic, ontology,
> reasoning, and computation built on top of the basic patterns that
> a child learns before the age of 3?
>

Hi, just to begin with:

Saunders Mac Lane, 1986. Mathematics: Form and Function. Springer Verlag
Light, P. & Butterworth, G. (eds.) - Context and Cognition - Hemel Hempstead, Harvester (1992)
George Lakoff and Rafael Núñez, 2000, Where Mathematics Comes From. Basic Books

But many of these ideas (as well as Piaget's) are built on top of pre-IIWW European naive physics and Gestalt movement.
Keep going!
Aldo

> SRF
>> When I try to use ontologies to do something useful in a computer
>> I frequently have to overcome the problem of converting data patterns
>> into symbols and symbols into actions.  There is still relatively
>> little research in how to bridge this semantic gap, if you consider
>> how important it is.
>
> The behaviorists in the US ignored that issue.  But European
> psychologists such as Piaget devoted decades of work to that topic.
> Seymour Papert, who collaborated on AI with Marvin Minsky at MIT,
> had spent several years working in Piaget's institute.
>
> From http://csjarchive.cogsci.rpi.edu/2008v32/7/HCOG_A_302167_O.pdf
>> The overall acquisition pattern — from relatively many mouth verbs,
>> to more hand verbs, to less bodily defined verbs — was unexpected and
>> tantalizing in its similarity to traditional Piagetian (Piaget, 1953)
>> descriptions of the developmental course of sensory-motor development
>> as infants first explore relations in their world.
>
> It's significant that researchers who are using the latest results
> from neuroscience find parallels to the stages that Piaget and his
> colleagues identified over half a century ago.
>
> MB
>> It seems to me that for ontologies that are relevant to businesses,
>> there must be a similar but separate grounding requirement, to that
>> of a child or other mammal.
>
> Piaget addressed those issues.  I certainly won't claim that he solved
> all the problems, but the framework is sound:
>
>  1. The pre-verbal sensorimotor stage is fundamental.  It begins
>     at birth and continues to develop for a lifetime.
>
>  2. A three-year-old child relates the words and patterns of language
>     to the sensorimotor schemata.  Older children and adults build on
>     that foundation, extend it, and make revisions where necessary.
>     But they never replace it.
>
>  3. Conceptual thought is a structural shift from the concrete
>     patterns to more abstract patterns.  Young children don't
>     understand metaphors, but the more abstract patterns are a
>     metaphorical application of the concrete patterns: The stock
>     market goes up or down.  Justice is served or thwarted.
>
> Following are some excerpts from one of Piaget's books.  For further
> discussion (with references) of related issues, see Sections 2 and 3
> of http://www.jfsowa.com/talks/goal.pdf .
>
> John
>
> ________________________________________________________________________
>
> Jean Piaget (1955) _The Construction of Reality in the Child_,
> translated by Margaret Cook, Routledge and Kegan Paul.
>
> Excerpts from the last chapter, "The Elaboration of the Universe":
>
> At the outset, two innovations place conceptual thought in opposition to
> sensorimotor intelligence and explain the difficulty of transition from
> one of these two forms of intellectual activity to the other. In the
> first place, sensorimotor intelligence seeks only practical adaptation,
> that is, it aims only at success or utilisation, whereas conceptual
> thought leads to knowledge as such and therefore yields to norms of
> truth. Even when the child explores a new object or studies the
> displacements he provokes, there is always in these kinds of
> sensorimotor assimilations, the concept of a practical result to be
> obtained.
>
> By virtue of the very fact that the child cannot translate his
> observations into a system of verbal judgments and reflexive concepts
> but can simply register them by means of sensorimotor schemata, that is,
> by outlining possible actions, there can be no question of attributing
> to him the capacity of arriving at pure proofs or judgments properly so
> called.  But it must be said that these judgments, if they were
> expressed in words, would be equivalent to something like, "one can do
> this with this object," "one could achieve this result," etc. In the
> behaviour patterns oriented by an actual goal, such as the discovery of
> new means through active experimentation or the invention of new means
> through mental combinations, the sole problem is to reach the desired
> goal.  Hence the only values involved are success or failure, and to the
> child it is not a matter of seeking a truth for itself or reflecting
> upon the relations which made it possible to obtain the desired result.
> It is therefore no exaggeration to say that sensorimotor intelligence is
> limited to desiring success or practical adaptation, whereas the
> function of verbal or conceptual thought is to know and state truths.
>
> There is a second difference between these two types of activity:
> sensorimotor intelligence is an adaptation of the individual to things
> or to the body of another person but without socialisation of the
> intellect as such; whereas conceptual thought is collective thought
> obeying common laws. Even when the baby imitates an intelligent act
> performed by someone else or understands, from a smile or an _expression_
> of displeasure, the intentions of another person, we still may not call
> this an exchange of thoughts leading to modification of those
> intentions. On the contrary, after speech has been acquired the
> socialisation of thought is revealed by the elaboration of concepts, of
> relations, and by the formation of rules. There is a structural evolution.
>
> It is precisely to the extent that verbal-conceptual thought is
> transformed by its collective nature that it becomes capable of proof
> and search for truth, in contradistinction to the practical character of
> the acts of sensorimotor intelligence and their search for success or
> satisfaction. It is by cooperation with another person that the mind
> arrives at verifying judgments, verification implying a presentation or
> an exchange and having in itself no meaning as regards individual
> activity. Whether conceptual thought is rational because it is social or
> vice versa, the interdependence of the search for truth and of
> socialisation seems to us undeniable.
>
> The adaptation of intelligence to these new realities, when speech and
> conceptual thought are superimposed on the sensorimotor plane, entails
> the reappearance of all the obstacles already overcome in the realm of
> action. That is why, despite the level reached by the intelligence in
> the fifth and sixth stages of its sensorimotor development, it does not
> appear to be rational at the outset, when it begins to be organised on
> the verbal-conceptual plane. On the contrary, it manifests a series of
> temporal displacements in comprehension and no longer only in extension,
> since in view of corresponding operations the child of a given age is
> less advanced on the verbal-conceptual plane than on the plane of action.
>
> In simpler terms, the child does not at first succeed in reflecting in
> words and concepts the procedures that he already knows how to carry out
> in acts.  If he cannot reflect them it is because, in order to adapt
> himself to the collective and conceptual plane on which his thought will
> henceforth move, he is obliged to repeat the work of coordination
> between assimilation and accommodation already accomplished in his
> sensorimotor adaptation anterior to the physical and practical universe.
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)