What is Behavior in Behavioral Economics?

Our Principle 1 says that “Choice is choice of behavior.” What exactly do we mean by “behavior”?

We would ordinarily expect to look to the field of psychology, the study of what people do and why they do it, to find an accepted definition of behavior, just we would look to physics to find an accepted definition of energy or momentum.  Unfortunately, we cannot do so.  In the field of psychology today there is no commonly accepted definition of behavior.

Perhaps the most commonly used definition of behavior in psychology is “an overt movement of the organism” [Bergner 2010], the definition that characterizes the behaviorist school. Cognitive psychology incorporates the same concept, but adds to it “processes by which the sensory input is transformed, reduced, elaborated, stored, recovered, and used” [Neisser, (1967) Cognitive psychology. New York: Appleton-Century-Crofts]. Functional psychology, a precursor to behaviorism, defined behavior as “purposeful activities to satisfy needs.”  (Though functionalism is not a current school of psychology, the functional concept of buying, an activity to satisfy needs, is central to the study of consumer behavior.) Freud took behavior to be the  expression of drives or instincts, themselves fundamentally neurological, but did not define behavior per se.

The inadequacy of the behavior-as-movement definitions (including the cognitive one) as a general foundation is illustrated by the afore-mentioned Friedman-Savage “paradox”.  What matters is buying the ticket or the insurance, not the physical motions of  doing so.  It is irrelevant whether the individual writes a check, takes cash out of a wallet, types and clicks on a computer, or makes a telephone call to buy the insurance.  Further, the fact that the individual is protected against damage after purchase is not representable by any set of statements involving physical quantities. That P “has” a policy is a legal relationship, not a physical one.  Somewhat similarly, for P to have the actual possibility of winning, she must have “possession” of the ticket, but that concept also cannot be defined with any possible set of statements involving physical quantities and physical relationships. Further, whether P engaged in “cognitive processing” in buying the policy, ticket, or both is irrelevant to the case; the case rests on the fact that individuals do buy both. Processing only enters the picture if a researcher insists on devising a process whose outcome would be the decision to buy. In other words, cognitive processes are a theoretical construct, not a part of the articulation of the facts of the case.

The point of a definition is to have an articulation of the subject matter and an identification of what is to be studied. It may be tempting to address the lack of a consensus definition of behavior by introducing a new one.  However, the failure to achieve a consensus definition in the more than 100 years the formal discipline of psychology has existed, and in the more general study of people and their behavior reaching back at least to the ancient Greeks, would seem to urge caution.  While traditional definitions of “behavior” have proven highly problematical, a different approach has proven much more fruitful. That approach is parametric analysis: rather than define what something is, identify the ways in which instances of that thing can vary, and treat each such dimension of variation as a parameter, whose values come from a known set of possibilities.

The classic example of the parametric formulation of a concept is its use to articulate the concept of color. We have no definition of color that identifies the set of all “things” that would be considered colors and excludes all that would not be. There is, however, a highly successful long-standing parameterization of color, used by artists and others who concerned with careful identification of colors, that identifies every color in terms of three values: brightness, hue, and saturation.  A specific color is identified by specifying three values, brightness, hue, and saturation, which can be written Color = (B, H, S); the parameters B, H, and S identify the ways in which colors may differ from one another (Gleitman, Fridlund, & Reisberg, 2004, pp. 190–191).

(Parenthetically, we note that the usual approach of defining terms is much more limited in applicability than is commonly recognized.  Every student of mathematical logic learns early that virtually all everyday concepts cannot be defined by a statement that identifies all the exemplars of the concept and excludes those that are not.  It is not, for example, possible to define “table” or “chair.”)

In the early 1960s, the psychologist Peter G. Ossorio used the method of parametric analysis to articulate the concept of human behavior. That articulation, “Intentional Action” (IA) has since become a central concept in the field of Descriptive Psychology and has been used by a number of researchers to address difficult research problems in a number of areas, including  economics [Jeffrey 2006, Jeffrey 2010], artificial intelligence [Jeffrey and Putman 1983, Jeffrey et. al. 1989, Jeffrey 1981], clinical psychology [Ossorio 2006, Bergner 2005], organizational psychology [Putman 1990], and modeling of social systems [Jeffrey 2007, Jeffrey 20003]. Just as specifying the values of B, H, and S completely specifies a particular color, specifying the IA parameters completely specifies a particular behavior, such as the one identified by saying “P bought a lottery ticket” or “P bought fire insurance.”

Principle 2: By “behavior,” we mean intentional action.

The intentional action formulation of the concept of behavior is:

Behavior = IA = <I, W, K, Kh, P, A, PC, S>, where:

  • I (Identity) identifies the individual whose behavior this is.
  • W (Want) identifies what the actor wants, that is, the state of affairs that the actor intends to bring about by engaging in the action.
  • K (Know) identifies what the actor takes to be the case and the distinctions he must make to carry out the behavior. “What the actor takes to be the case” includes what are informally called perceptions, beliefs, verified fact, and more generally whatever the actor can, in principle, act on.
  • Kh (Know-how) identifies competencies the actor must have, to do this behavior.
  • P (Performance) identifies the actual process the actor performs to carry out the behavior.
  • A (Achievement) identifies the state of affairs that the behavior actually achieves.
  • PC (Personal characteristics) identifies the characteristics of the person this  behavior is an expression of, when there are such: traits, attitudes, interests, styles, abilities, knowledge, values (motivational priorities), states, and capacities.
  • S (Significance) identifies the larger behavior this one is part of, i.e., what the actor is doing by doing this.

Several of these parameters require some elaboration. Before that, we illustrate the analysis with the two behaviors we have previously seen: buying fire insurance and buying a lottery ticket.

The parametric formulation of P buying fire insurance is:

  • I: P
  • W: to be protected against financial loss due to a house fire
  • K: the distinction between fire vs. something else, fire insurance vs. something else, protection from damage and something else, the actual possibility of serious loss from fire, the amount she must pay, the coverage provided, various fire insurance providers, various policies
  • Kh: competencies required to seek out and purchase fire insurance and to assess one provider and policy vs. another
  • P: all procedures involved in purchasing the insurance
  • A: P is protected against loss from fire, as specified in the purchased policy
  • PC: We do not have enough information to conclude that this behavior is an expression of a particular characteristic of P.  The behavior is congruent with P being a prudent person; it is also congruent with P seeing a neighbor’s house burn and acting “in a panic” – i.e., in that panic state (which is a PC).  If additional examples of either kind of behavior are observed, a PC conclusion would be warranted.
  • S: by buying fire insurance, P is protecting herself against catastrophic harm, which is itself a behavior.

Buying the lottery ticket is formulated as:

  • I: P
  • W: to have the actual possibility of winning $100M.
  • K: the distinction between a lottery and something else, a lottery ticket vs. something else, the dollar cost of the ticket, the effective cost of the ticket, the possibility of winning $100M; distinction between what P’s life is like now vs. what would be if she won $100M
  • Kh: competencies necessary to find a location selling lottery tickets and actually making transaction
  • P: the procedure involved in purchasing the ticket.
  • A: P has the actual possibility of winning $100M.
  • PC: We do not have enough information to conclude that this behavior is  an expression of a particular characteristic of P. However, the behavior is congruent with P being in a state of financial anxiety, or having at least some interest in gambling
  • S: makes possible a major improvement in her life circumstances

Elaborations on the parameters

Extensive discussion of these parameters may be found in Bergner [2010] and particularly Ossorio [2006], but certain aspects of them need some clarification here, for the purposes of this paper.

The state of affairs P wants (W parameter) to bring about is also P’s reason for acting.  Colloquially, we say P “has reason enough to act.” W is a state of affairs, but a state of affairs maybe comprised of several other, constituent, states of affairs, related in various ways. Thus, P may want a piece of pie and a friendly conversation. The resulting state of affairs may be that P is engaging in the behavior, the phenomenon of intrinsic behavior, i.e., the well-recognized fact that in many cases what the person wants is to be doing the behavior, as the above-mentioned examples of playing chess and going shopping illustrate.  W is not, therefore, the traditional concept of incentives.

“What the actor takes to be the case” is intentionally non-committal as to traditional “truth-value.” It includes what are informally called perceptions, beliefs, verified fact, and more generally whatever the actor can, in principle, act on. None of the reformulations and technical elaborations, such as cognitive structures and processes, belief networks, etc. found in cognitive models and artificial intelligence are part of this concept. In particular, the K parameter and its values are not probabilistic concepts; we do not assume that probabilities are associated with what the actor takes to be so, nor that that any rules involving probability calculations, such as Bayesian calculations, are or are not appropriate models of the K parameter values.

Competencies and skills are not the same concept.  Competencies are a parameter of behavior, whereas having a skill is a Personal Characteristic, attributable on the basis of having observed the individual engaging in a sufficient range of behaviors, in a wide enough range of circumstances. Skills are acquired by having had an learning history involving behaviors requiring the relevant competencies.

The value of A is a single state of affairs, as with W. Paradigmatically, the value of A is the same as the value of W, i.e., in the paradigm case P achieves what he sets out to achieve.  However, as noted above, a state of affairs may be comprised of several other SAs, so it may be that what is achieved is a complex state of affairs comprised of other related ones, which may need addressing – side effects, unintended consequences, or simple errors. What is achieved is this single, overall, state of affairs A, with various constituents.  The formal structure of states of affairs is greatly elaborated Ossorio [2005]-WAH and Jeffrey 2010a] For example, in the case of buying fire insurance, A is having protection and the state colloquially referred to as “peace of mind,” a fact that is universally recognized and acted on by sellers of home insurance.

While the values of the W, K, Kh, and A parameters in the above examples are ordinary English phrases, the Intentional Action specification of a behavior is not simply a discursive exposition arranged in a particular way.  The common language locutions specifying the parameter values are identifiers of states of affairs (or a process, in the case of the P parameter). Once so identified, these locutions may be used as formal names in the formal state of affairs system [Ossorio 2005, Jeffrey 2010a, Jeffrey 2010, Jeffrey 2007].  In systems using formal state of affairs representation, additional formal names may be used in addition, such a PF100, ProtectedAgainstLoss (a style found in modern computer programming languages), W100M, Z33.1, etc.  As with behavior itself, most states of affairs cannot be formally defined, but they can be articulated, using the approach of identifying a state of affairs by name and further specifying it by specifying its constituents and their relationships, again by name. This methodology is discussed at length by Ossorio [2005], and formulated in a mathematical style in Jeffrey [2010a].  The central concept is that an actor acts to bring about a state of affairs, which can be identified by a name, a common language expression that serves to identify it, whether or not it can be defined in the classic mathematical sense.

In contrast with virtually all other current conceptualizations, W and K do not denote anything “internal,” such as “internal states,” “cognitive structures,” “mental objects or processes,” etc.  They also do not denote any physiological state, structure, or process, including neurological or hormonal ones.  K, for example, identifies what the actor takes to be the case and the distinctions he is acting on, not a “state of mind,” thought process, mental construct, etc.

The parameters provide the logical link necessary for doing non-reductive neuroeconomics and other neurophysiology.  They formally specify how discovered neurophysiological processes and states are related to what the actor does in the real world, i.e., his behavior.  For example, one may investigate what neurological processes and structures are involved in recognizing instances of concepts (K parameter), what neurological structures and processes are involved when an individual is presented with an opportunity to get something they value (W parameter), etc.  It is not the case, however, that these processes and structures are constitutive of the behavior or any aspect of it, other than the P parameter, the steps involved (which, as articulated below, can be described in terms of sub-processes, sub-sub-processes, etc., in the usual way).

The Kh parameter is responsive to the need to address the fact that persons commonly do things – that is, engage in behaviors – for which we cannot specify a procedure, and in many cases for which no procedure has ever been specified, behaviors which we customarily say they “learned how” or “know how”  to do.  A major league baseball player can routinely hit a baseball over the fence in batting practice; a trained physician can diagnose many diseases with a high degree of accuracy; some investors can select a basket of securities whose collective performance far exceeds that of a randomly selected basket. The values of the Kh parameter are the states of affairs the actor has the competence to bring about, but not via a procedure. Physicians do not identify a disease or condition by engaging in a procedure (though they may engage in procedures, such as getting test results, before they make the diagnosis); financial analysts do not select securities by engaging in a procedure (though they may engage in procedures, such as studying a company and an industry, before they make the selection).

The traditional formulation of this kind of fact has been to assume that there is an “unconscious process” doing the calculation.  This methodological assumption is seen clearly in the writing of Kahneman and Tversky [1974, 1986, 2000], in which they explicitly describe decisions in terms of Bayesian probability calculations and flaws in how those calculations are carried out.  It is almost universal, for example, to refer to choice as a process [Khaneman and Tversky 2000, p. 32]. The intentional action formulation deliberately refrains from making such assumptions, taking the most conservative approach possible, namely, incorporating only what is observable about a behavior, rather than including hypothesized “mental processes.”

The P parameter, the observable “behavioral episode” or performance, itself has extensive structure, specified by a Process Description [Ossorio 2006]:

  • The Stages of the process, a task description of the steps involved in a process
  • The Options for one or more stages: alternative processes that produce the same outcome of a Stage
  • Constraints on the occurrence of a Stage or Option, in terms of the occurrence of some other Stage or Option, or on some object (which may be a person)  involved in the process having some attribute.
  • The logical roles in the process, which Ossorio terms Elements
  • The actual individuals that may serve as each of the Elements
  • The Eligibilities of each individual to be each Element, which include governing rules
  • Versions, the actual sets of Stages (and Options for a Stage) that may occur and be considered to be occurrences of the process.

The Eligibilities parameter encodes the precise change in state of affairs that is accomplished by buying a lottery ticket: a ticket-holder is eligible to win the lottery.

An important point for our purposes here is that what actually occurs is the instantiated version of a process.  Thus, a complete specification of an actual behavior is specified by specifying the values of the IA parameters and the instantiated version of the P parameter.

Personal Characteristics are a parametric analysis of the ways in which persons – economic actors – can vary. The direct connection to behavior is that 1) sometimes a behavior primarily reflects one or more characteristics of the actor, rather than the circumstances, and 2) observed patterns of a kind of behavior give a describer reason to ascribe a characteristic to the person.  A cautious investor, for example, is one who repeatedly and regularly chooses the cautious investment option, when presented with the choice.

Where does money fit?

Our purpose here is to formulate a new paradigm for economics, not merely to discuss behavior in general, and we therefore must identify where money fits in the conceptual framework. We will begin our next post with this discussion.

This entry was posted in New Paradigm and tagged , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s