Paul J. M. Jorion
Comments in Behavioral and Brain Sciences
Comment on Byrne, R.W. & Russon, A.E. Learning by imitation: A hierarchical approach.
Byrne & Russon's text displays all the difficulties involved in approaching imitation in a non- methodological behaviourist perspective. Their conceptual apparatus is grounded in a mix of introspection and folk psychology. Their distinction between action level and program level imitation falters on goal imputation for sequential acts. An alternative gradient descent model is propounded: behaviour can be modelled as a frustration/satisfaction gradient descent within the animal's potentiality-space as defined by knowledge, inventiveness and surrounding universe.
A methodological behaviourist model for imitation (1998)
Byrne & Russon's Learning by imitation displays fully the difficulties involved in any approach to imitation other than methodological behaviourism. The initial flaw resides in their conceptual apparatus for imitation, grounded in a mix of introspection and folk psychology in a classical three-step process.
Step 1: I rationalise my own behaviour in purposive terms borrowed from folk psychology as inscribed within everyday language, i.e. I generate for my own autobiographical use a post hoc discourse where my consciousness assigns itself goals (final causes) then accomplishes them,
Step 2: I expand the scope of the template to rationalise the observed behaviour of my fellow human beings (the supporting evidence remains though that of my initial introspection),
Step 3: I further expand the scope of the scheme, this time to cover animal behaviour (the same observation applies as for step 2).
Once reached step 3, I am stuck with the encumbrance of what appeared innocuous enough at inception: consciousness in the driver's seat. I am faced in the case of animals with the impossible resultant task of sorting out voluntary acts from involuntary, purposive sequential acts from accidental; in addition I am being forced for plausibility's sake to jettison consciousness in its causal role of behaviour somewhere arbitrarily down the evolutionary ladder between apes and insects, etc.
Consequently, Byrne and Russon's distinction between action level imitation and program level imitation falters on goal imputation for motor act sequences which is grounded on what amounts to the "compassionate introspection" of the researcher,. For instance, in the authors' own words: "In some cases (of orangutan behaviour), their goals were clearly those of the humans, for whatever reason" (§ 2.6). "Clearly" to whom one may ask, if the reason turns out to be obscure? Similarly, "... her behaviour made sense only with this goal in mind" (§ 2.6). Is "last attempt at making sense" strong enough support for goal imputation? And so on.
For a number of years now I have propounded for behaviour a gradient descent model eschewing the pitfalls of goal imputation (Jorion 1990: 94-97; 1994: 94-98: 1997: 3-4). Within such framework consistent with methodological behaviourism, any sequence of animal (and human) behaviour can be modelled as a frustration/satisfaction gradient descent within the animal's (human being's) potentiality-space as determined by its knowledge, inventiveness and current state of its surrounding universe.
I will illustrate this on examples from the text. A human being reaches satisfaction (relaxation of frustration) when achieving to light a stick dipped in kerosene. An orangutan reaches satisfaction when human bystanders burst with laughter, caused by her failing to light a stick dipped in kerosene despite astute "aping" (§2.6). "Failing" to ignite the stick belongs here fully to the ape's successful strategy (pace Byrne & Russon: "Despite failure to effectively execute the entire program of fire-making..." §2.6) as her satisfaction results from the laughter, the fire-making being foreign to her monkey business.
Similarly in the other example of orangutan behaviour, when satisfaction is not obtained through successfully washing clothes (apes wear no clothes) but through having the "obligingly shrieking staff" jump into the water (§2.6). Byrne & Russon's choice of the word "obligingly" betrays their being fully aware that the ape is not seeking satisfaction through washing clothes but through interacting with the human actors on the scene. That once the staff is in the water the orangutan then diligently goes about washing the clothes, is not problematic: it is a perfect example of stimulus enhancement as defined by the authors "the tendency to pay attention to, or aim responses towards, a particular place or objects in the environment after observing a conspecific's (here human beings) actions at that place or in conjunction with those objects" (§1.2).
Such relaxation towards satisfaction can obtain throughout any technique - or any combination of techniques - known to the individual: from on-the-spot invention down to reproduction of recorded sequences of action level imitation. There is here but a single principle at work: the frustration/satisfaction gradient. At each stage of the gradient descent the individual proceeds from the local point currently attained in the potentiality-space so as to make the rate of frustration reduction the highest possible: or in visual terms, the choice made at every local point attained is that of the path offering the steepest slope.
The model is of general application. In particular it explains away a number of anomalies mentioned in Byrne & Russon's Learning by imitation:
1. Action level imitation is intrinsically comical to any onlooker, i.e. it has high social interaction satisfaction potential and is used to this purpose by great apes in their intercourse with men. This feature of action level imitation does not preclude it from being also a possible component of a problem-solving strategy. This excludes however that there is anything like "copying behaviour for its own sake" (§2.6): imitation is in every instance a strategy for attaining some satisfaction, either instrumental or social.
2. About Abranavel & Gingold's bathing a teddy-bear task (1985), Byrne & Russon remind us that "Any actions unrelated to the job in hand tended to be missed out in the children's imitations" (§2.4). These "unrelated actions" do not contribute to satisfaction (frustration reduction) and are consequently left out from the sequential acts leading to frustration relaxation.
3. More importantly, there is no need to devise a theory that explains how memory "keeps track of where to return to [...] as the problem is 'unpacked' into component sub-goals" (§2.2.) as satisfaction obtains through a gradient descent. Sub-goal imputation is a post hoc reading of the descent path. Alleged sub-goals single out singularities in the potentiality space landscape where slopes are at their gentlest.
Jorion, Paul (1990) Principes des systèmes intelligents, Collection Sciences Cognitives, Paris: Masson, summary in English:http://cogprints.soton.ac.uk/abs/comp/199806039
Jorion, Paul (1994) L'intelligence artificielle: au confluent des neurosciences et de l'informatique, Lekton, vol.IV, 2, 1994: 85-114;http://cogprints.soton.ac.uk/abs/comp/199807008
Jorion, Paul (1997) Ce qui fait encore cruellement défaut à l'intelligence artificielle, Informations In Cognito, N° 7: 1-4;http://cogprints.soton.ac.uk/abs/comp/199807007