A JANETIAN APPROACH TO PALAEO-PSYCHOLOGY
Muséum National d’Histoire Naturelle de Paris
e-mail : email@example.com
According to P. Janet’s psycho-philosophy, psychology is the science of « conducts ». Defined many
years before Watson’s article, this at once enlarged behaviorism keeps behaviors for animals only. A
human conduct is a behaviour plus typical human mental activities such as consciousness, reasoning,
ideas, which ought not be discarded by scientific psychology. This theory is the one needed to address
complex human activities during palaeolithic periods, as an alternative both to evolutionary psychology
and cognitive psychology.
A simulated experiment
Psychology of goal oriented actions is a valuable tool provider in the attempt to model palaeolithic man
behaviour and action categorization, because both experimental psychology and paleo-ethnology aim at
collecting action data, sequences or problem solving competences. These two sets of data are equally
compatible with direct (psychology) or simulated (paleo-ethnology) experiments. So far paleo-ethnology
hasn't explored the usefulness of psychology methods. We address here the methodology of such an
attempt and give preliminary results.
Categorization : a trans-specific mental mechanism
Humans and animals behave adaptively in response to a novel stimulus, because it resembles others for
which the appropriate response is already known. Such an adaptation expresses an ability to categorize. In
the absence of categorization, each object or event would be perceived as unique and generalizations
would be impossible. Therefore, it is not surprising to find categorical abilities in various animal species
(J. Vauclair, The cognitive animal, 2002). While language depends on categorization, categorization does
not depend on language (Herrnstein, 1984), and as a fundamental aspect of information processing,
categorization is hence a first choice mental mechanism to investigate in past human species.
Birds and primates
Since the pioneering work of Herrnstein and Loveland (1964), psychologists have repeatedly
demonstrated that birds categorize stimulus classes containing variable instances. The pigeons'
extraordinary perceptual capacities render it an ideal species for investigating categorization (Lea 1984,
Wasserman 1991, Mackintosh 1995). A useful general framework for the investigation of these behaviors
was provided by Herrnstein (1990), who described categorization abilities in animals in five levels of
increasing abstractness. See Zayan / Vauclair (1998) and Thompson / Oden (2000) for reviews. The
evidence for capacities to perform the first three levels of categorization is large for several animal
species: macaques, cebus, baboons, chimpanzees etc. (J. Vauclair, The cognitive animal, 2002).
Humans and prehistoric hominids
Two common kinds of models exist to describe the structure of knowledge : Collins & Quillian's (1969,
1974) and Rosch's (1973, 1975, 1978). These two models view categories structures as a hierarchical tree.
But at the end of the nineteenth century, P. Janet suggested that action be nor a characteristic of the acting
man, nor one of the objects being handled. Actions are relations between humans and objects : analyzing
them is to analyze both parts equally. One of the consequence is that objects can be viewed as the actions
they afford. Starting from similar ideas, J.F. Richard (1994,1997) proposed a cognitive model of human
categorization. This is most interesting for archaeology, whose very scope lies in reconstructing humans
activities, namely what they did with the objects found.
Unfortunately, as far as paleolithic intelligence is concerned, most studies define their own vocabulary
and method, usually using current words, ascribing them various meanings (Belfer-Cohen & al. 1994,
Saillot & al. 2000). No debate can hence be fruitfull because of a lack of common basis (Wynn, 2002).
Experimental psychology allows to understand and model paleolithic man behaviour (Saillot, 2001), for
any human species or paleolithic period (Saillot & al. 2002).
The Galois lattice : A visual representation of categories
After the pioneering work of Collins and Quillian (1969) who designed a visual representation of
concepts maps in memory (semantic networks), the Galois lattice has been proposed by Barbut and
Monjardet (1970) as a convenient tool to achieve concepts classification. Galois lattices consist in
grouping objects into classes that materialise concepts. Individual objects are discriminated according to
the properties they have in common. It was popularized by Wille (1982, 1992) who proposed to consider
the nodes as concepts, with their extension and intension. There are currently many methods for the
construction of such lattices (Chein 1969, Bordat 1986, Ganter 1988, Missikoff 1989).
Application : semantic network of a simulated experiment on palaeolithic categorization
Subsistence activities during palaeolithic is the set of human conducts regarding the use of the
environment. Its study relies on the analysis of butchery marks on animal bones. If the data about
subsistance actions is rich enough, a palethnologic conclusion can be drawn. Archeozoologic studies lead
to the definition of operational chains of acquisition and processing of animals by prehistoric man (Patou-
Mathis 1993). It is possible to reduce the processing of an animal to a sequence of technical butchery
steps. A big mammal holds many ressources for food and other purposes : it can be considered as a source
of raw material. Butchery operations lead to the collection of these raw materials, which are objects that
can be associated with different goals and procedures by paleolithic man. According to P. Janet’s point of
view, these goals or procedures are the objects properties, hence they allow the construction of
categorization lattices. Following P. Janet’s requirements, we suggest that the research for conducts
causes is a methodological flaw : description is what we aim at.