Behaviorism


Behaviorist Arguments

Criticisms of Behaviorist Arguments

Behaviorist Responses to Criticisms

Observation: The Case against Introspection

Introspection--i.e., internal observation of one's own consciousness--is to be rejected because consciousness is not an objective object to be observed.  BUT that argument either
     
  1. begs the question of what is objective, or 
  2. is based on metaphysical notions of objectivity. 
Introspection is to be rejected because it is especially prone to error.  BUT introspection, when done properly, can be as objective as any other kind of observation. 
Introspection is to be rejected because it is unreliable. Introspection is unreliable because the verbal community is unable to match exactly discrimination of verbal self-reports of internal stimuli (e.g., "I feel hungry") and discrimination of private internal stimuli (e.g., hunger pangs). Therefore, the link between introspective verbal reports and internal stimuli is unreliable. (Skinner [1986], for one, holds this position.) 
There is a public-private distinction.

 Introspection is to be rejected because introspective evidence and observations are private, i.e., available only to one observer, and are thus not objective.

     
  1. Private = subjective. 
  2. Public = objective. 
BUT introspective observations, like UFO reports, can be verified indirectly from
     
  1. evidence about the subject's prior and subsequent behavior, 
  2. the subject's accuracy in reporting other kinds of observations, and 
  3. reports of other introspectors. 
For example, "lie" items on self-report questionnaires (e.g., "I read every editorial in the newspaper every day") indicate that a subject has a certain kind of response set unlike that of others, and is probably responding inaccurately in order to appear in a better light. 
YES, BUT introspective observations, unlike UFO reports, are not publically verifiable in principle. It is impossible to establish a correlation of eating behavior with an internal state like hunger (as opposed to just self-reports of hunger). 
There is no public-private distinction.

 All observation is private: Knowledge of the external world is really an inference from the immediate but private percept. 

BUT that means there is no reason not to trust private introspection.  NO, BECAUSE science is collaborative. Public (not private, introspective) behavior affords the most agreement and communication among observers. Therefore, in place of the subjective vs. objective distinction, the following methodological distinction should be made.
     
  1. Private = subjective. 
  2. Public = intersubjective. 

The Behavioral Data Language: Selection of a Domain for Psychology 

The is an a priori distinction between behavior and physiology.

 Molar Behaviorism: A science of behavior autonomous from physiology is possible and desirable.

     
  1. Molar behaviorism, but not physiological psychology, leads to easy prediction and control, partly because behavioral variables are observable. 
  2. Concern with physiology leads to lack of concern for behavior. 
  3. Physiological psychology finds a proximate cause, but leaves it unexplained, opening the door to magical explanations in terms of agents and wills. 
BUT the boundary between behavioral and physiological events is difficult to define. 
The is an a posteriori distinction between behavior and physiology.

 Behavioral research leads to the discovery of laws, some of which cover more than behavior simply conceived. "Behavior" can be defined as whatever conforms to these laws. These laws are then organized by behavioral theories. 

The Behavioral Data Language: Selection of a Descriptive Language for Observational Reports 

There can be purely objective descriptions.

 Although theory influences observation, this influence can be overcome by observing more carefully, thereby leading to an objective database. 

BUT all observation is theory-laden YES, BUT to evaluate the thesis of theory-ladenness requires either
     
  1. theory-neutral facts, or 
  2. theory-laden facts from another theoretical orientation, and 
  3. if the theory-ladenness thesis is itself a theory, then it will dictate observations that confirm itself. 
There cannot be purely objective descriptions.

 There is a continuum from more to less theory-laden observational descriptions. Intersubjective agreement on descriptions indicates an acceptable degree of "objectivity." 

BUT intersubjective agreement may indicate only that observers share a common theoretical framework or paradigm.  YES, BUT that is acceptable. The science of behavior can proceed as normal science within its own theoretical framework or paradigm. 
Purpose does NOT belong among the descriptive qualities of behavior.
     
  1. Intentional, purposive descriptions are highly interpretive, and therefore do not allow intersubjective consensus. 
  2. Purposive qualities should always be explained in terms of more basic properties of behavior. 
Purpose DOES belong among the descriptive qualities of behavior.
     
  1. Intentional, purposive qualities of behavior can be attributed to organisms on the basis of objective criteria, and therefore allow intersubjective consensus. 
  2. Purposive qualities are emergent, and therefore cannot be explained in terms of more basic qualities of behavior. 

Theoretical Concepts

Theoretical vocabulary should be restricted to operationally defined concepts.

 Theoretical concepts are to be linked to the data language through operational definitions. 

BUT operationism does not provide suitable criteria for individuation of concepts. Any two things can be considered similar or different in an infinite variety of ways. (Cf. the discussion of problems of description in sociobiology [Lewontin, 1979].) One cannot know when to
     
  1. ignore differences and consider two things similar, or 
  2. ignore similarities and consider two things different. 
There is no unique shorthand summary of observations. 
YES, BUT
     
  1. distinctions are actually drawn only after the fact, in such a way as to maximize simplicity and comprehensiveness. 
  2. Even an operational definition is subject to change after the fact. 
Theoretical vocabulary should NOT be restricted to operationally defined concepts, but should include INTERVENING VARIABLES.

 Use of intervening variables should be allowed, but such variables are only labels for observed relationships, they are not unobserved entities or events. There is no unique set of valid intervening variables, and they have no unique representation. 

BUT
     
  1. intervening variables are not truly explanatory. Black-box, input-output laws may allow prediction, but not explanation. 
  2. Hypothetical constructs should be allowed because they lead to scientific progress. The history of science shows that unobservable hypothethical constructs have had the properties imputed to them when they were later able to be observed. 
NO, BECAUSE black-box theories conform to the deductive nomological method of explanation (Hempel & Oppenheim, 1948): descriptions of behavior are deduced from general laws plus initial conditions. 
Theoretical vocabulary should NOT be restricted to operationally defined concepts, but should include HYPOTHETICAL CONSTRUCTS.

 Use of hypothetical constructs should be allowed, and such constructs refer to unobserved entities or events, much as in physics. 

BUT psychology is unlike physics in that
     
  1. in psychology theoretical terms are not introduced through a postulate set, 
  2. psychologists bring folk psychology concepts into their work, and 
  3. psychology is a less well-developed science. 
HYPOTHETICAL CONSTRUCTS (cont.)

 Use of hypothetical constructs should be allowed, because hypothetical constructs bridge the gap between psychology and physiology. 

BUT
     
  1. hypothetical constructs may be no better for theoretical reduction than intervening variables, and 
  2. physiological mechanisms are only conventions in the same sense that intervening variables are. 

Theorizing

Within behaviorism there are two modes of theorizing. Both presume that there exists a rigorous division between theory and observations.
     
  1. Hull's hypothetico-deductive method: A theory allows the deduction of theorems about observables, and these theorems can be tested experimentally. A theory that is confirmed by repeated tests allows prediction and control. (There is nothing so practical as a good theory.) 
  2. Skinner's experimental analysis of behavior: Research should be conducted not to test theorems, but to find orderliness in behavior. The search for orderliness is intuitive, rather than structured by scientific method. A theory is a formulation that uses a small number of terms to explain a large number of facts. 
BUT
     
  1. observation is not theory-independent, it is theory-laden. Therefore, observation can neither test a theory (Hull) nor broaden a theory (Skinner). 
  2. Data do not dictate theory; rather

  3. (a) theory determines what is taken as fact,
    (b) theory cannot always adjust to fit contradictory data, and
    (c) sociological factors, rather than data, often determine a theory's acceptance or rejection. 
Theory-ladenness of observations just implies that the science of behavior will progress as normal science, i.e., linearly and continuously (until a scientific revolution overthrows behaviorism and replaces it with a new paradigm). 

S-R Psychology

No major behaviorist holds that behavior is reflexive, i.e., that a stimulus is necessary and sufficient for a response. 
The Non-Cognitive Learning Approach

 Environmental events that do not elicit a particular movement can come to do so through learning. Thus, a particular stimulus is not a necessary condition for a particular response. Learning consists of the association of stimuli and responses. 

The Cognitive Learning Approach

 BUT learning should be described as the acquisition of knowledge. 

NO, BECAUSE then the relationship between knowledge and behavior remains unspecified. In theories postulating cognitive maps or observational learning, behavior is derived from the theory by assuming that the organism will behave appropriately given its knowledge. Such assumptions are not acceptably objective. 
Events inside the organism, as well as outside, can be considered "stimuli." BUT Skinner, for one, holds that it is a mistake to suppose that there are internal stimuli (Evans, 1968, p. 21). 
Observed behavior is the result of integration of many reflexes. Therefore, it may have properties not possessed by any individual reflex. 
Operant responses are emitted, rather than elicited by stimuli. Thus, the S-R reflex thesis can be reduced to the claim that all behavior is functionally dependent upon the environment.  BUT the environment and behavior are causally interdependent.  YES, reciprocal determination, in which organisms alter the environment that controls them, is incorporated into many behaviorist theories. 

The Organization of Behavior: Purpose

Behaviorists roundly reject teleological explanations, without denying that behavior has purposive characteristics (e.g., persistence, flexibility). These characteristics are accounted for non-teleologically in the three ways detailed below. 
Purpose is a state of the organism that determines the relationship between stimuli and responses. Purpose is a state variable that is brought about by certain motivational variables and that increases the probability of a certain class of responses. 
Purpose can be explained as stimuli that are maintained until eliminated by a goal-response. Behavior shows persistence because different goal-responses will continue to be emitted until the stimuli disappear. Behavior shows flexibility when new goal-responses succeed in eliminating the stimuli. 
Goal-directedness is only apparent, not real. Responses occur not to achieve future goals, but because they were reinforced in the past. Behavior shows persistence because of intermittent reinforcement. Behavior shows flexibility because many different kinds of responses may be reinforced, and thus an organism may possess a large repertoire of responses. 
(This table is adapted from Zuriff, 1986.)


References

Evans, R. (1968). B. F. Skinner: The man and his ideas. New York: Dutton.

 Hempel, C., & Oppenheim, P. (1948). Studies in the logic of explanation. Philosophy of Science, 15, 135-175.

 Lewontin, R. C. (1979). Sociobiology as an adaptationist program. Behavioral Science, 24, 5-14.

 Skinner, B. F. (1986). Is it behaviorism? Behavioral and Brain Sciences, 9, 716.

 Zuriff, G. E. (1986). Précis of Behaviorism: A conceptual reconstruction. Behavioral and Brain Sciences, 9, 687-724.