Structuring decision problems and the ‘bias heuristic’

This paper examines our understanding of the decomposition of immediate acts when structuring decision problems. Seven different types of uncertainties are identified, and four of these are shown to be taken explicitly into account in models within the province of decision theory, described in terms of four interlocking systems interfaced with semantic memory (a core act-event system, and systems buffering utilities, probabilities and events, respectively). Requisite decision modeling is shown to require that the remaining three types of uncertainty (procedural uncertainty; how the decision maker will feel about subsequent acts; agency for changing subsequent states of the world) are also resolved. Methods for ‘fixing’ structure are discussed in terms of aiming at a common understanding about the ‘small world’ in which a decision problem is located. Difficulties in resolving uncertainties in doing this are described. An alternative approach, common in studies invoking ‘behavioural decision theory’ is contrasted: imposing structure, assuming common understanding. The latter approach is shown to involve (i) the ‘naturalisation’ of the small world in which the decision problem is located, and (ii) the utilisation of normative models as ‘ideal types’, leading to the use of the ‘bias’ argument in discussing subjects' performance in decision tasks. Using this argument reflexively, the operation of the ‘bias heuristic’ is identified in a survey of published papers referencing this approach to the study of decision making. Effects identified are: availability of tasks, subjects and explanations; representativeness of findings; and anchoring and adjustment of explanations. Implications for practice are discussed throughout the paper.     

Pavlovian processes and response competition as determinants of avoidance response-prevention effects

Two experiments investigated the facilitation of avoidance extinction by exposure to lengthy (5-sec) shock during avoidance response prevention. In Experiment 1, animals exposed to light only or to light-shock pairings during response prevention showed equal facilitation of extinction relative to shock-only animals or to animals receiving no response prevention. Preshock rearing, directly antagonistic to the avoidance response, developed for shocked animals during response prevention and persisted during extinction for light-shock animals. Immediately before extinction, half of each group was permitted a single escape from a light-shock compound by means of the response previously required for avoidance. The only effect was upon the extinction performance of light-shock animals. Rearing was eliminated and extinction responding increased to a level far above that for any of the other animals. Experiment 2 demonstrated that the shock-only treatment affected the extinction performance and rearing of nonescape and escape animals in a manner entirely equivalent to the effects of the light-shock treatment of Experiment 1, provided stimulus conditions (light absent) were the same for all experimental phases. Thus, lengthy shock during avoidance response prevention simultaneously leads to the acquisition of competing behavior and enhances control by a warning signal or contextual stimuli over the avoidance response. Implications for the CS-only response-prevention treatment and the transfer of aversive control are discussed.     

Natural, unnatural, and supernatural concepts

Five experiments investigating information-processing consequences of concept naturalness and unnaturalness are reported. In particular, an attempt is made to examine the separate and joint effects of simplicity and entrenchment upon reasoning with natural and unnatural concepts. The experiments employed four variants of a concept-selection task in which subjects characterized the state of an object at some future time on the basis of information about the object at that time and at some earlier time. In all five experiments, both simplicity and entrenchment affected ease of information processing. The last two experiments showed that the effect of simplicity was greater than that of entrenchment. An informationprocessing model of task performance provided a very good account of the latency and error data. Scores from the concept-selection task correlated with scores from a set of psychometric inductive-reasoning tests; indeed, the correlations were higher than those usually found between cognitive task performance and psychometrically measured intelligence. Processing of nonentrenched concepts may provide a useful means for studying the nature of intelligent performance.     

Aphasic adults' use of heuristic and structural linguistic cues for sentence analysis

Eight aphasic adults with relatively preserved auditory comprehension and 12 normal adults were tested for their ability to analyze sentences in which the functional relations between the verb and nouns was either directly reflected by word order expectancies, or marked by a functor or non-NVN word order. The subjects' task was to designate the agent of each sentence. Results indicated that aphasics utilize word order expectancies but are impaired in their ability to exploit functors and particularly, word order markers for sentence analysis. Results are discussed in terms of aphasics' use of heuristic versus structural linguistic cues for interpreting sentences.     

Blocking in successive differential conditioning: Prior acquisition of control by internal cues blocks the acquisition of control by brightness

In each of four experiments the acquisition of discriminative stimulus control by brightness cues (black vs white runway) in a successive go/no-go instrumental discrimination was blocked in groups given prior discrimination training with internal (reward-produced and intertrial interval-related) cues as relevant discriminanda and brightness cues irrelevant. The blocking effects obtained here in instrumental conditioning were substantial and in most cases complete. Blocking occurred whether brightness in Phase I varied across trials but was uncorrelated with reinforcement (varied-irrelevant, V-I, condition) or whether all Phase 1 trials occurred to a single value on the brightness dimension (constant-irrelevant, C-I, condition) which then served as a redundant S+ cue in Phase 2 while the previously unexperienced brightness cue was added to the S- stimulus compound, or vice versa. Blocking in the V-I condition was shown not to be due simply to nondifferential reinforcement of brightness in Phase 1, but to depend on the prior acquisition of discriminative stimulus control by internal cues. As in Pavlovian conditioning, blocking here was an increasing function of amount of prior conditioning to the blocking stimulus. The results encourage the prospect that the procedures used here can be developed into a viable instrumental conditioning companion to the Pavlovian procedures now used almost exclusively to study blocking.     

Self-punitive behavior in humans: Effects of a self-fulfilling prophecy

Human subjects were exposed to contingencies which programmed aversive tones (100 db). Two types of contingencies were employed: self-confirming (i.e., self-fulfilling prophecies), in which the aversive tone was occasioned by the prediction it was about to occur; and self-disconfirming, in which the tone was probable when subjects predicted it would not occur. Experiments 1 and 2 used a modified classical conditioning paradigm, and demonstrated that a self-confirming contingency maintained reliable self-punitive responding, i.e., subjects consistently predicted and therefore obtained tones on every trial. Subjects in Experiment 3 were instructed to express predictions continuously throughout four sessions to ensure adequate sampling of the various predictions. Subjects exposed to a self-disconfirming contnngency reliably evidenced awareness of the contingency in effect (judged by answers on a postexperimental questionnaire), whereas subjects exposed to a self-confirming contingency failed to show effective avoidance behavior or contingency awareness. Experiment 4 investigated free-operant self-punitive behavior, utilizing a single prediction response button, which subjects depressed repeatedly. Subjects were exposed to either periodic or aperiodic punishment schedules over as many as four sessions. In general, more persistent self-punitive responding was found in the groups receiving periodic punishment. The results from the four experiments show that self-confirming contingencies can effectively prolong self-punitive responding in human subjects. The findings are consistent with a blocking interpretation of self-punitive behavior, which asserts that when an aversive event is already predicted by stimuli in the situation (including temporal cues), the association between a response and punishment is impaired, and self-punitive responding is likely to be maintained. An integration of human and animal self-punitive research is proposed.     

On the matrix formulation of Kaiser's varimax criterion

The author provides a full-fledged matrix derivation of Sherin's matrix formulation of Kaiser's varimax criterion. He uses matrix differential calculus in conjunction with the Hadamard (or Schur) matrix product. Two results on Hadamard products are presented.     

Nontransitive measurable utility

A new theory of preferences under risk is presented that does not use the transitivity and independence axioms of the von Neumann-Morgenstern linear utility theory. Utilities in the new theory are unique up to a similarity transformation (ratio scale measurement). They key to this generalization of the traditional linear theory lies in its representation of binary preferences by a bivariate rather than univariate real valued function. Linear theory obtains a linear function u on a set P of probability measures for which u(p) > u(q) if and only if p is preferred to q. The new theory obtains a skew-symmetric bilinear function φ on P × P for which φ(p, q) > 0 if and only if p is preferred to q. Continuity, dominance, and symmetry axioms are shown to be necessary and sufficient for the new representation.     

Intermodal perception of expressive behaviors by human infants

Four experiments demonstrate that infants of 5 and 7 months can detect information that is invariant across the acoustic and optic presentations of a single affective expression. Infants were presented simultaneously with two filmed facial expressions accompanied by a single vocal expression characteristic of one of the facial expressions. The infants increased their looking time to a facial expression when it was sound-specified, as compared to when that filmed expression was projected silently. Even when synchrony relations were disrupted, infants looked proportionately longer to the film that was sound-specified, indicating that some factor other than temporal synchrony guided the infants' looking behavior. When infants viewed the filmed facial expressions either in a normal orientation or upside-down, those infants viewing the facial expressions in the normal orientation looked appropriately, while those viewing the inverted films did not. These findings support the view that infants are sensitive to amodal, potentially meaningful invariant relations in expressive behaviors. These results are discussed in the context of J. J. Gibson's theory of affordances.