This page is dedicated to the memory of
Albert Paul Weiss (1879-1931), an early behaviorist
and the most important early voice expressing concern about the need to distinguish between
bodily movements and accomplishments, presenting
what we here call "the incommensurability
problem."
Included at this page is a link to a biographical essay, written by Professor Robert H. Wozniak at Bryn Mawr, the editor of the recent reprint of Weiss' 1925 classic, A Theoretical Basis of Human Behavior. a bibliography of Weiss' work, three obituaries (pending), and a discussion of Weiss' argument regarding the incommensurability problem.
The little that is known of Weiss' life is to be found principally in three obituaries. Permission to reprint these obituaries has been obtained and all three will be added to this WebSite shortly. The three are:
Bloomfield,
Elliott,
Renshaw,
the incommensurability problem:From a scientific perspective, it is simply the case that important properties of a system understood at a higher level cannot always be calculated directly from properties at a lower level. The different descriptions are then said to be incommensurable.
The important thing about incommensurability is that, where it occurs, no amount of theorizing will make syntactic reductions possible. Another solution must be found. Thorough knowledge of neural activity might, in principle, allow us to predict bodily movements. Weiss (1924; 1925) demonstrates that descriptions of bodily movements and actions are incommensurable.
Let's look at what Weiss had to say:
Behavioristic psychology occupies an intermediate position [between the physical sciences and the social sciences], on the one hand investigating the effects of physical conditions on sensorimotor functions, and on the other the effects of sensorimotor function on social [interaction]. Its specific contribution to general science I regard as that of bridging the gap between the physical and the social sciences.... Thus there arise two criteria with respect to which human movements may be classified: (a) as neuromuscular effects of preceding movements, (b) as neuromuscular causes of subsequent movements. I have differentiated these classifications by the terms biophysical and biosocial. I shall use the term biophysical for the [precedent] neuromuscular... phase of the human response, and the term biosocial when a response is classified with respect to its [subsequent] stimulus character either for other individuals or for the same individual at a later time....If the study of human behavior is to achieve a scientific status..., both the biophysical and biosocial properties must be studied. I wish to direct especial attention to the fact that biophysical and biosocial do not refer to two aspects of the same thing, say the type of sensorimotor organization. The classifications which include responses that are biosocially similar need not show any biophysical similarity [and vice versa].
Weiss (1924, p.42-44), see also Weiss (1925, p.55-56).
What Weiss (1924, 1925) seems to say (I am doing a contemporary reinterpretation here) is that there is a central principle that must guide the development of any theory in a genuinely behavioral psychology. Any psychological theory must cope with the fact that every behavioral event may be classified in at least two ways, first in terms of the biophysical causes as bodily movements and second in terms of the effects upon environmental circumstance as accomplishments. Further, Weiss notes that the two taxonomies, that of movements and that of accomplishments, are incommensurate.
A few examples will make this clear. The first issue is whether or not, knowing the accomplishment, we could determine the bodily movement (and thus the neural activity). There are innumerable examples to demonstrate that this is not the case. We may open the door with our hand or with our shoulder or our foot if we are carrying a package. An object may be grasped with the right hand or the left. Etc. Etc. Etc. The converse case is not so readily seen. To adapt an example from Weiss (1922, pp.331-332), the bodily movements in signing one's name on two different occasions might, in principle, be identical, but, if one signature is on a check and the other is on a love note, the accomplishments in the two cases are quite different. Lee (1986, p.172) gives and eloquent and thoroughly non-linguistic example. The bodily movements in watering one's garden and in hosing down a driveway may be identical. Thus, we might know, or be able to predict, the neurophysiological activity in detail and thereby be able to predict the bodily movements and yet still have no idea what was being accomplished. (Waismann, 1965, pt. II, ch. VIII, p.158, offers another colorful if slightly gruesome example: the distinction between fencing and dueling.)
Much later, Austin (1962) pointed out that even a single particular instance of a behavioral event may have multiple interpretations in terms of accomplishment with his example of John crooking his finger, pulling the trigger, firing the gun, shooting the man, killing the man, killing the King, signalling the start of the revolution, all descriptions of a single event. Within Philosophy, this is known as the problem of the individuation of action; to what category does a unique individual act belong. Weiss is careful, as a monist, to say that the distinction between movement and accomplishment is strictly one of classification. For Weiss, the incommensurability of bodily movements and accomplishments demanded a methodology that allowed behavioral events to be classified in both ways. We may thus call this claim the requirement of double individuation. Thus, the incommensurability problem is a specific case of the more general problem of the individuation of action. In general, it is difficult to identify the appropriate category for a particular action. In specific, the categorial taxonomies specified by neurophysiological similarities (and thus the taxonomies to which act-individuations in any neurophysiologically based theory would appeal) is incommensurate with the taxonomies of accomplishments used to describe the empirical regularities which any psychological theory is called upon to explain.
Now, Weiss does not really say why he takes this feature of behavior to be central to behavioral psychology, but we may suppose that he does so, in part, to forestall the critiques of the mentalists who, as in the case of critiques of Watson, construed behaviorism as neglecting these problems. For instance, Blanshard (1939) makes the inability of any reductionist behaviorism, such as Watson's, to deal with multiple meanings of biophysically identical linguistic vocalizations as a telling and final demolition of the possibility of any behaviorism ever working. (Obviously, Blanshard had not read Weiss.) Weiss agreed with the critics that the fact that pronouncing the sentence "This is the place where the son's rays meet," could be biophysically identical to the sounds produced by a proud papa in pointing out his male children's cattle ranch ("This is the place where the son's raise meat.") was a genuine problem for any behavioral psychology. Weiss insists that this problem must be dealt with head on.
In order to address the justification of Weiss' claim as to the centrality of the requirement of double individuation, we need to look ahead fifteen years to Guthrie (1940). Guthrie provides a truly eloquent justification, which I will summarize:
The data of psychology, with certain important exceptions (such as some Classical conditioning) is given in terms of relations between circumstances and accomplishments and changes in those relations. Now, suppose we have data regarding learning, which is to say, changes in the relations between circumstances and accomplishments over time. In terms of physical events, we may divide the sequence of events into three contiguous portions. The sequence from circumstances to sensory inputs. The sequence from sensory inputs to bodily movments. And the sequence from bodily movements to accomplishments. During learning, the relations governing the first and third sequences obviously do not change. Therefore, it is changing relations between sensory inputs and bodily movements that cause the changes in the relations between circumstances and accomplishments that we call learning.
Now, suppose that neurophysiology were to provide us with a comprehensive theory of nervous function. Presumably, such a theory would contain in it everything we need to know about learning. Of course, as Guthrie says so eloquently, since neither the environment of light rays reflected off nearby objects nor the objects manipulated by our effectors are innervated, a theory of nervous function can only tell us about the relations betweem sensory inputs and bodily movements. Now, it may be, as Guthrie suggests, that the pattern of relations between sensory inputs and bodily movements may be homomorphic to the observed regularities in the relations between circmstances and accomplishments, but given the incommensurability noted by Weiss, this homomorphism cannot be guaranteed a priori.
Let us assume for a moment that a neurophysiologically based theory can act as a true theory of behavior in the same sense that Maxwell's kinetic theory gives a true theory of Boyle's and Charles' Laws, etc., in that it explains the underlying regularities in terms unlike the observational terms (i.e., what Nagel calls heterogeneous reduction). If so, we would need a general method for deriving predictions in terms of circumstances and accomplishments given neurophysiologically-based predictions in terms of sensory inputs and bodily movements. Given that Weiss entitles his book "A Theoretical Basis for Human Behavior," we may suppose that it is the necessity for understanding double individuation as a prerequisite to reducing behavior to neurophysiology along the lines of Maxwell's kinetic theory of gases or Physical Chemistry's reduction of chemical valence to electron orbitals that justifies Weiss' claim that double individuation is central to the study of behavior.
In short, a case can be made that Wiess believed that double individuation was prerequisite to the development of true theory in psychology.
Of course, we could follow Skinner (1950) and advocate that only empirical laws (similar to Boyle's Law and Charles' Law) are the province of behavioral psychology. Skinner argues compellingly that the incommensurability problem does not arise in the development of empirical law. But Skinner (in Catania & Harnad, 1988, p.128, pp.469-470) admitted that his solution was a stop-gap. Ultimately, the explanations provided in terms of behavioral law must be buttressed by biologically-based accounts. At that point, the full force of Weiss' dilemma returns.
Austin,
Blanshard,
Catania
Guthrie,
Lee,
Skinner,
Waismann,
Weiss,
Weiss,
Weiss,