William T. Moran, Lever Brothers Company, New York
[ to cite ]:
William T. Moran (1974) ,"Consumenoid", in NA - Advances in Consumer Research Volume 01, eds. Scott Ward and Peter Wright, Ann Abor, MI : Association for Consumer Research, Pages: 257-267.

Advances in Consumer Research Volume 1, 1974    Pages 257-267


William T. Moran, Lever Brothers Company, New York

Consumenoid I is a theory of human behavior which is inclusive and which at this point, is quite complete in broad, conceptual outline, fairly well along in operational definitions, and in a nascent state with respect to computer programming. CI, which is Consumenoid's nickname, is a model of a single person's behavior mechanism; it is not a macro-probabilistic nor analytic model. Thus, interactive behavior and aggregate population behavior must be built up through simulation. Whereas, the processes within an individual CI are largely deterministic, population behavior would be probabilistic due to chance distributions of individual characteristics and experiences.

Considering the comprehensive behaviors which CI is intended to explain, the number of mechanisms or variables is parsimonious. Because CI is not a simple, linear heuristic--its having numerous interactions, parallel processes and feed-backs--these few mechanisms produce a richness of functions and a wealthy variety of behavior. As a result, despite the basic determinism of CI it is virtually impossible to trace back from any specific resultant behavior to an analytic interpretation of causes. The same is true of individual human behavior. The value of a model, of course, is that the experimenter can begin with alternative causes and then observe specific behaviors with the privilege of re-setting real time and circumstance.

The structurally lean and functionally complex nature of CI poses a dilemma with respect to assessment of subsystems in the model akin to the difficulty of observing the function of a biological organism when the parts are separated. The organism doesn't necessarily behave the same way separately as it does together.

We do not expect that the parameters we observe from manipulation of our partially complete computer model will continue to be exhibited when the rest of the interactions are incorporated. It is for this reason that we want to stress the basic conceptual framework before we examine the current state of the computerized version of Consumenoid and the parameters which it reveals. Familiarity with earlier papers describing CI (6,7,9,10,13,14) will prove helpful in understanding its constructs, but the following overview of the major elements or "parts" will provide a general background.


An individual is presumed to behave in response to information from the perceptual universe. For CI the perceptual universe is three-dimensional: comprised of wants, action alternatives and stimuli. Any phenomenon may be described in any one or more of these dimensions.


The information being received from perceptual universe and perceived at any time constitute the phenomena from which the individual identifies one or more perceived situations. The signals being received vary in probability of being perceived according to:

a. variability

b. association with other signals already perceived (the other signals may be on the same dimension or on either of the other two).

A perceived situation is a higher order of organization than random, unassociated signals and, therefore, represents a reduction in perceived entropy in the perceptual universe.

Perceived situations are defined as probabilistic sets of signals classified in the three dimensions. The probabilities arise from the individual's prior experience with their concurrence and, as some expected concurrent signals are present and perceived, the probability of a previously experienced situation is "recognized."

Inasmuch as no situation is exactly replicated in the real world, "recognition" is a probabilistic estimate. As the perceptual field and the signals vary from moment to moment, the set outlines of the probabilistic clusters are constantly shifting in lesser or greater degree, much as the penumbra and umbra of shadows, altering the recognizability of the situation or changing to another perceived situation of a higher order probability.

A perceived situation is a matrix of signals on the three dimensions: wants, action alternatives, stimuli. Once enough concurrent signals are perceived to raise the possibility of a known situation, the individual actively scans for associated signals which also are expected. In this manner weak or ambiguous signals are perceived, identified and incorporated into the probability that the hypothesized situation exists.

In a perceived dinner situation with bread on the table butter would normally be expected, and the individual may actively search for this missing action alternative which meets a want which "belongs" in the situation. Similarly, the perception of an action alternative in a situation sets off an internal scanning of associated wants to determine whether or not that "belonging" want signal is perceived.

Whether a signal is perceived as an action alternative or not is a function of whether the signal previously has been associated in a similar situation as an action alternative. Signals which are seen as action alternatives in one situation are not seen as action alternatives in some other. For example, if one suddenly were asked to identify a weapon nearby, one might he nonplussed and perceive none. Whereas, if annoyed by a fly, one might quickly perceive the newspaper as a weapon. In other situations a belt, an ashtray, a pencil might be seen as weapon alternatives.

Where no satisfactory action alternatives are perceived, an individual may conceive of action alternatives out of past personal or vicarious associations and defer action until the option can be exercised. Thus, an individual may "create" action alternatives, or, similarly, intentionally or unintentionally may alter the perceived situation by an oblique action which eliminates the need for the previously sought action alternative. All intentions are not carried out.

Stimuli signals may be resolved into wants or action alternatives by the context of the perceived situation. Faced with ambiguous or dissonant signals, the individual seeks additional signals to confirm or deny alternative situation possibilities. For example, if one were to wake up and find oneself standing in a ballroom in the midst of a party, holding a screwdriver, the question would come to mind whether one were a member of the party or, instead, a workman. It would not be unusual if, under such dissonant circumstances, one looked to see how he was dressed in order to decide whether one were holding an inappropriate object or else standing in the wrong place.

In another case to illustrate the interchangeability of the three perceptual dimensions as accesses to situation perception: one might hear chimes and, depending upon one's experiences and other signals, might consider the possibility that someone is at the front door or that Church service is being held or that it is mealtime. A scan of one's wants might evoke the information that one is hungry. The individual then searches for situation completion and establishes whether or not it is mealtime by reference to a clock, the sudden notice of the aroma of cooking, etc.

It often has been observed that awareness of hunger will suppress if the "mealtime" situation is perceived at a time when it is not possible to eat. Later, despite the presumed continuation of the biological need to satisfy the hunger, the sensation of hunger diminishes or disappears under the "distraction" of other situations with which the hunger want is not associated. Similarly, one of the problems of obesity is that to some people the sight or smell of food is translated into a mealtime situation and the hunger want is evoked irrespective of the recency and quantity of food consumed.

From this discussion of the perceptual process in the Consumenoid theory it is apparent that the identification of situation requires experience on the part of the individual. A newborn CI enters a perceptual world of chaos unless supplied with artificial experience and associations to speed up maturation. Also, the benefits of experience will vary according to the association formation ability and the memory of the individual CI. These individually variable attributes--experience, memory, and information processing ability--play important roles in the different individuals.

It should also be noted that CI theory calls for signals to be organized by the individual at different levels of complexity by clustering and labelling such as might be described by factor analytic techniques. Also, it is presumed that some wants signals vary in intensity in accordance with biological clocks as well as in response to actions. The intensity of a signal such as a want can grow to the point that it overwhelms other signals and, itself, becomes the perceived situation. Otherwise, the awareness of a want is a function of whether or not it is evoked through association with the perceived situation. Also, in accordance with George Miller's observation in "The Magic Number Seven, Plus or Minus Two," CI theory presumes that there is a limit to how much information an individual can handle at a time. The notion of perceived situations permits wants and other signal associations to be evoked or suppressed and for the evocation or awareness of an additional signal to tend to suppress some other signal less salient to the perceived situation. In this manner the potential signal repertoire in an individual's perceptual universe can be very large without requiring that more than a few signals or signal clusters be perceived at any time.


In Consumenoid theory an individual behaves in contemplation of forecasts of outcomes from alternative perceived action possibilities. The forecasts are probabilistic; that is, there are error ranges. There are three classes of error. There is real system error arising from the degree of stability in the relationships between actions and their effects on the individual's wants structure. There is the forecasting error introduced by the breadth of the individual's experiences, processing ability for the formation of associations from experience, classification of situation and memory. And, quite a different error range, is the difference between individuals in their confidence in their ability to forecast; that is, the error ranges they assign to their forecasts.

It is easy to visualize that the behavior of otherwise identical individuals in identical situations would vary if either their forecasting ability or their confidence in their forecasting ability differed. Also, depending upon the stability of the system, the outcomes could vary in response to identical behavior--thus, through feedback, altering the probabilities of behavior choices in similar situations in the future. In order for these and other functions of CI to operate it is necessary that the model be operated with a directional time stream so that events can be sequenced into a chronological order.


The notion of planning horizons plays a key role in Consumenoid theory. Even in the case of perfect information and perfect logic the optimum decision can (and often will) differ according to the planning horizon employed. This phenomenon is familiar to financial investment planning and to management science.

As a result, individuals with the same intelligence and information will behave differently if they employ different planning horizons. It is possible that the failure of educational approaches to the problems of anti-social behavior may be due less to differences in information or in intelligence than in the planning horizons employed. There is very little literature on how businesses or individuals or societies should select planning horizons.

It is a premise in CI that individuals differ in their planning horizon repertoires: some employing a wide variety and others only a narrow selection. The majority of individuals, it is believed, employ different planning horizons for different situations. It is speculated that the choice of a planning horizon will be affected by: the number of action alternatives and wants evoked by the perceived situation, the level of confidence in the forecasts, the extent to which confidence in the forecasts varies over longer and wider planning horizons, and the extent to which the individual believes it can effect post hoc reversals of outcomes.

It is conceivable, for example, in societies where there is little confidence in the stability of relationships between actions and outcomes that it would appear logical to employ very short and narrow planning horizons. Such a basis for behavior would be expected to result in very low levels of cooperation between individuals, high orders of immediate self-gratification, and little investment in future productivity. Similarly, children and other unsophisticated individuals, due to limited experience, can evoke relatively limited situation sets with simple wants, action alternatives and expectations. They might behave, however, with a confidence in outcomes which could prove dangerous to themselves or others. The value of vicarious experience, such as through reading, observation of others, or other means of communication can reduce personal danger, speed up sophistication and provide continuity to society.

The CI planning horizons are conceived as having both temporal and spacial dimensions. That is, a planning horizon can extend from the immediate present into the future--even beyond one's lifetime. Also, a planning horizon can be narrow or broad with respect to the number of wants, action alternatives, stimuli with which a currently contemplated action might interact (be associated) now or in the future--i.e., the complexity of the perceived situation, or system size.

Because so little information has been gleaned from the professional literature on the genesis, population distribution and employment of alternative planning horizons in human behavior, Professors Donald L. Kanter and John Gutman of the University of Southern California Graduate School of Business have undertaken a research program to develop such information. The results of their work will be made available to the Institute for The Quality of Life for the benefit of Consumenoid applications to practical social problems.


As there is a planning horizon, so is there an experience horizon. Feed-back of outcomes from past actions (direct or vicarious) provides a network of possible associations useful for forecasting. The length and breadth of the experience horizon that an individual can and does employ will influence the complexity of the system (situation) evoked, the accuracy of and confidence in forecasts and, thereby, the amount of search and behavior variation displayed on future occasions as the individual seeks to achieve satisfaction.


Finally, we come to that part of Consumenoid which accounts for why it behaves at all. There always exists the alternative behavior of doing nothing in a given situation. Sometimes individuals do nothing relevant to a given situation and simply move on (or are moved) to a different situation. Real time passes and, with its passage, the penumbras and umbras of perceived situations shift and regroup into new shadows, new clusters of probable situations with new stimuli evoking new wants, new action alternatives.

But sometimes we act. Why? To accomplish what end?

Behavior seems so contradictory. A Robinson Crusoe lands on a smooth, deserted beach and is impelled to draw designs in the sand or to dig holes and make piles. Someone else arriving at a popular resort beach at the end of the day is greeted by hundreds of sand castles and is tempted to kick them down. Children build up columns of blocks only to delight in knocking them over. Some people build more than they destroy; others destroy more than they build; most of us do some of each on different occasions. Some people conserve; others take great risks. Some fight for life under great privation; others are self-destructive. There are masochists and hedonists. Some exploit; others create. Some cooperate; others compete. Some avoid obstacles; others seek out obstacles to be overcome. But rarely are we the same in all situations.

The Consumenoid theory to account for this kaleidoscope of behavior from individual to individual and from situation to situation is that the underlying motive force for all is to maximize the-cumulative absolute change in the degree of structure in the perceived situation over the planning horizon employed.

The CI presumption is that the prime mover is the desire to effect as much change as possible in relation to the perceived cost to and resources of the individual (talent, energy and degrees of freedom). This estimate of the ability to change perceived structure through action is made over the length and breadth of the planning horizon so that an individual who is employing a long planning horizon will forego present gratification if he forecasts that, by so doing, the cumulative gratification (cumulative change in the situational structure) over successive actions will thus be greater.

The measure of structure employed in CI is the entropy of the three-dimensional situation matrix. Because the three dimensions are mutually dependent, by definition, through association in the experience horizon the individual might conceive of herself as behaving in relation to any one of them alone: e.g., affecting the action alternatives, the stimuli or the wants, without necessarily thinking in terms of the interaction effects on the others. Similarly, from a measurement standpoint, the change in entropy of any of the dimensions serves as a practical surrogate for the change in the entropy of the entire matrix.

Because the objective is to maximize absolute entropy change, increases in wants constitute gratification in the same way as do decreases in wants. Thus, periods of rest alternate with periods of activity, episodes of creativity alternate with periods of dissipation. The individual who perceives himself confronted by chaos will seek to create shape and structure. The individual who feels surrounded by such a highly structured society that he cannot conceive of proliferating the structure through his puny talents and resources will feel compelled to pull it all down around him (our doctrineless revolutionaries) or else to slip into catalepsy if he has no confidence in his ability to effect that course either.

Whereas Consumenoid theory holds that the objective of behavior is to maximize cumulative absolute entropy change in the perceptual universe, it does not suppose that individuals are aware of this complex mechanism of gratification. The effect of an act on the wants structure will be to alter the structure and, thus, the entropy so that the next behavior which will be most gratifying will be different from what it might have been if a different action had been taken. The internal, unconscious adjustment of the individual will be in accordance with the change in entropy.

The conscious decision rules of the individual, however, will be some far simpler set of rules of thumb which (if the individual is a successful survivor) will roughly approximate the same results over time as does the actual biological process of cumulative entropy change. Efforts to rationalize and to articulate the motives underlying behavior will be vague, inconsistent, and conflicting as will the decision rules themselves.

Thus, there will be some inevitable differences between actual outcomes and perceived outcomes which, should they prove cumulative, will create great anxiety and dissatisfaction with life. A persistent discrepancy between actual and perceived outcomes in similar situations will preclude needed revision of forecasts and will result in what will appear to others to be compulsive behavior.

Outcomes can and will differ from forecasted outcomes and, to the extent that these discrepancies are perceived, the individual will revise the forecasts and possibly also the evoked wants and/or action alternatives in future similar situations. In the same fashion planning horizons and forecast confidence are subject to change through experience.

Because of the time dimension of the planning horizon outcomes of actions can appear to unfold gradually as, of course, they often actually do. Because of the space dimension of the planning horizon wants, action alternatives, stimuli are affected (some in minor degree) which also are affected by other actions in other situations, by biological clocks, by exogenous forces. The knowledge of outcomes, therefore, is never complete nor certain. It becomes a question, then, what we mean by "satisfaction." In Consumenoid theory degree of satisfaction is established through the exercise of post hoc forecasts of the outcome of the past action. To the degree that successive post hoc forecasts continue to repeat the original forecast as more and more information comes in, then "satisfaction" is felt.


So much for the underlying constructs of Consumenoid. With a partially complete computer program we have begun to explore the parameters of behavior which some of the "parts" of CI produce. For example, we have wondered about the effect on simple choice behavior in a single, repetitive situation of planning horizons of various lengths.

It should be emphasized that, whereas earlier in this paper the individual Consumenoid model is described as "largely deterministic," the model as currently programmed is entirely deterministic; there are no Monte Carlo processes whatsoever.

The experimental runs of the CI computer program conducted to date enable us to observe the behavioral consequences only of a few of the processes: (l) the basic entropy change drive with respect to wants structure and action alternatives, (2) the planning horizon, and (3) the experience horizon. Both the planning and experience horizons are with respect only to length of time--i.e. future purchase opportunities. The breadths of the horizons are constant, reflecting only two of the three situational dimensions: action alternatives (always the same four brands) and wants (always the same five wants).

Also, in the current runs there are no forecasting errors (CI gets exactly the satisfactions it predicts) and outcomes are fully and accurately known prior to the next purchase occasion. The biological clocks re-set the wants after satisfaction back to exactly the same levels as before the prior purchase. And the current CI has full forecasting confidence. Each occasion produces a single purchase and consumption of a single unit from which the individual receives expected utility regardless of who else might do the actual consuming and also receive utility.

Given the above constraints on the realization of Consumenoid theory, it is not expected that the observed purchase behavior will be naturalistic. In this manner it is, however, possible to conduct sensitivity tests of the planning and experience horizons and to observe the contribution of the basic entropy change drive under the extreme conditions of full confidence--errorless forecasting and full awareness of outcomes--with no other intervening behaviors nor other factors altering the wants structure from purchase occasion to purchase occasion and no new information about the brands, their availability nor stimulus salience.

Successive sensitivity tests of a similar nature will enable us to accumulate an understanding of the behavioral implications of each of the functions and of their interactions. It will then be possible to construct a simulated population of Consumenoids and to study the behavioral consequences in comparison with empirical aggregated data.

In the current experimental series, CI was presented with the following conditions (Table 1). Forecast (and achieved) wants satisfactions for each brand are binary (1: present; 0: absent). Wants levels are normalized and can range from O - 1.0. A brand which satisfies a want reduces the level to 0.

Because the experience horizon reflects behavior, prior experience was generated by running the program in order to avoid the forcing of a possibly uncharacteristic behavior history. The brand history for a given experience horizon was employed to determine the brand history for longer experience horizons. For example, the distribution of the last twenty purchases (of a series of forty purchases) resulting from an experience horizon of one was used to determine the distribution of prior purchases for an experience horizon of three, of three for five, etc.

The reported behavior for each condition, as in the examples (Tables 2,3) are only for purchases 21-40. As can be seen from Table 2, with an experience horizon of ten purchases (entirely of Brand A) and with a planning horizon of one, CI continues to purchase only that brand which provides the maximum wants reduction. Such a result is implicit in standard attribute perception models and in "rational" man theory on the assumption that wants reduction is the basic drive.

Table 3 however illustrates the consequences of a change in the planning horizon to five purchase occasions ahead (and of the altered brand purchase history available to the experience horizon as a result of generating the experience by going from an experience horizon of one to three with the same planning horizon). Here we see a substantial amount of variety seeking despite the total determinism (additional runs under the same conditions will produce the identical pattern). It will also be noted that despite the fact that Brand C provides the least wants reduction, it obtains a substantial share of this individual's business apparently because it provides a combination of satisfactions not otherwise obtainable. The amount of variety seeking is extreme and undoubtedly due to the fact that so few functions are operating in this experimental series. The variety index, 0 in Table 1 and 120 in Table 2, is a simple expression constructed from the sum of the deviations in the distributions of shares of purchases from random expectations (maximum variety). The index ranges from 0 (minimum variety) to 150 (maximum variety). It would appear to matter a great deal to purchase behavior if one believes one is going to "only go around once in this world".

In Table 4 the results of the experimental series described in Table 1 are summarized. The summed results indicate the behavior is more sensitive to the planning horizon than to the experience horizon. However a very short experience horizon has an inhibiting effect on the variety seeking with longer planning horizons.

These are just the simple interactions, and more intensive analysis of the larger number of runs generated by the experimental series remains to be done. In addition there are numerous investigations to be carried out in order to reveal the behavioral parameters produced by Consumenoid theory. Early experiments include the consequences of the addition of a fifth brand which is unique versus one which offers the same satisfactions as an existing brand. Anyone interested in conducting experiments through the Consumenoid model should contact the author or the Institute for the Quality of Life, 300 Park Avenue, New York, New York, 10010.




RUN #518


RUN #569




Esienberger, R. Explanation of rewards that do not reduce tissue needs. Psychological Bulletin, 1972, 77, 319-339.

Fiske, D. and Maddi, S. Functions of Varied Experience. Homewood, Ill.: Dorsey Press, 1961.

Foerster, H.V. What is memory that it may have hindsight and foresight as well. Artorga, February, March, April, 1969.

Garner, W.R. and McGill, W.J. The relation between information and variance analyses. Psychometrika, September, 1956.

Hughes, R. The decline and fall of the avant-garde. Time, December 18, 1972.

Light, M.L. Consumenoid I: an explanation and illustration. Paper presented at the 80th Annual Convention of the American Psychological Association, Honolulu, 1972.

Longman, K.A. Applications of Consumenoid I. Paper presented at the 80th Annual Convention of the American Psychological Association, Honolulu, 1972.

Maloney, J.C. Spare parts bin for a new social psychology. Contemporary Psychology, 17, 1972, a review of Howard, J.A. and Sheth, J.N., The Theory of Buyer Behavior.

Moran, W.T. Saints and sinners. Address on the meaning of the Consumenoid model of human behavior, presented at the 80th Annual Convention of the American Psychology Association, Honolulu, 1972.

Moran, W.T. Methods of psychology in marketing. Paper presented to the European Society for Opinion and Marketing Research, Maidenhead, England, 1973.

Rosen, S. The future of hindsight and foresight. The Wall Street Journal, February 19. 1972.

Sackett, G.P., Keith-Lee, P. and Treat, R. Food vs. perceptual complexity as rewards for rats previously subjected to sensory deprivation. Science, 1963, 141, 518-520.

Starr, M.L. Management: A Modern Approach. New York: Harcourt, Brace, Javanovich, 1971, 591-683.

Starr, M L. Micro and macro aspects of behavior modeling: amplification of the entropy change notion underlying CI's behavior. Paper presented at the 80th Annual Convention of the American Psychological Association, Honolulu, 1972.

Teilhard de Chardin, P. The Future of Man. London: Colins, 1964 (translated by Norman Denney).

Watt, Kenneth. A moveable (disappearing) feast. Saturday Review of the Sciences, February, 1973.

Wicker, Tom. The wrong model. The New York Times, July 27, 1972.