Some new fundamental considerations of variety-seeking behavior

Abstract

This article deals with living systems at the organism level. Through aggregation we obtain the behavior of groups or classes of persons as they exercise choice perogatives within society. The study focuses on subsystems which process information with major emphasis on the associator, the memory, and the decider. Fundamental advances in modeling human choice behavior (at the level of an individual human being) have occurred as a result of computer simulation capabilities for testing various hypotheses about sustained loyalty over a sequence of decisions, e.g., repeating brand choice. The models reflect the fact that choice behavior can be viewed as an information process. One of the prime descriptions of information sequences is the Shannon entropy measure which has been used by social scientists for about the same length of time as computers have existed. This paper proposes a unique modification of the entropy level measure to explain switching patterns of loyalty such as: one choice alternative is repeatedly selected, but the consistency is punctuated by occasional switching to others. The modified measure relates to changes in entropy. Specifically, the most successful hypothesis for explaining observed loyalty is that the individual strives to maximize the cumulative absolute change of entropy, i.e., max Σ ‖ ΔHt ‖ = max Σ ‖ Ht − Ht−1 ‖. This dynamic viewpoint necessitates consideration of the effects of planning horizons (how far in the future is T?) and experience horizons (how far back does memory extend?). The latter is reflected in the calculation of Ho. Unequal costs of alternatives (such as brands) are shown to produce stronger loyalty effects for the less expensive choices. This leads to the postulate that individuals perceive lower costs for choices that change entropy and higher costs for choices that provide minimal changes. Thus the best description of a variety-seeking system over time is not measured by the net change in its entropy (Σt ΔHt), but by the absolute change (Σt ‖ ΔHt ‖). Zipf's empirical studies (e.g., concerning word lengths) are cited in support of these notions. The aggregation of simulations of choices by individual human beings based on the max Σ ‖ H ‖ hypothesis presents challenging opportunities to study and explain social behavior.