An assessment of response bias for thesame-different task: Implications for the single-interval task

Two auditory amplitude discrimination experiments were conducted using thesame-different experimental design. Observer bias was manipulated, in the first experiment, by varying payoff matrices and, in the second, by varying prior probabilities of signal presentation. Five levels of bias manipulation and four levels of difficulty were employed in each experiment. Each observer received all combinations of bias manipulation and difficulty, but with only one of these combinations within each block of trials. Nine indices of bias were assessed by simultaneously fitting isosensitivity and isobias functions to the data and by fitting isobias functions only. Although none of the indices tested provided an exceptionally good fit to the data, two indices stood out from the rest. These werec _sd ^* andc _i, indices with isobias contours similar in shape to those for thec index derived from theyes-no task.     

Indices of response bias in the same-different experiment

The same-different experiment is popular for assessing perceptual performance, including the performance of people with neuropsychological deficits. Although the measurement of accuracy with this experiment is now well understood, the measurement of response bias remains problematic. Reformulating the decision space for the experiment can yield new bias indices that are analogous to, but numerically different from, those of the more familiar yes-no experiment. Isobias curves that show how hit rate covaries with false alarm rate for constant bias but varying accuracy are presented for eight indices, and best-fitting parameters of the isobias functions are determined for a set of experimental data. The theoretical status of the bias indices and their relation to other formulations are reviewed.     

Indices of response bias in the same-different experiment

Thesame-different experiment is popular for assessing perceptual performance, including the performance of people with neuropsychological deficits. Although the measurement of accuracy with this experiment is now well understood, the measurement of response bias remains problematic. Reformulating the decision space for the experiment can yield new bias indices that are analogous to, but numerically different from, those of the more familiar yesno experiment. Isobias curves that show how hit rate covaries with false alarm rate for constant bias but varying accuracy are presented for eight indices, and best-fitting parameters of the isobias functions are determined for a set of experimental data. The theoretical status of the bias indices and their relation to other formulations are reviewed.     

Reinforcer control and human signal-detection performance

Eight humans participated in a two-choice signal-detection task in which stimulus disparity was varied over four levels. Two procedures arranged asymmetrical numbers of reinforcers received for correct left- and right-key responses (the reinforcer ratio). The controlled procedure ensured that the obtained reinforcer ratio remained constant over changes in stimulus disparity, irrespective of subjects' performances. In the uncontrolled procedure, the asymmetrical reinforcer ratio could covary with subjects' performances. The receiver operating characteristic (ROC) patterns obtained from the controlled procedure approximated isobias functions predicted by criterion location measures of bias. The uncontrolled procedure produced variable ROC patterns that were somewhat like the isobias predictions made by likelihood ratio measures of bias; however, the obtained reinforcer ratio became more extreme as discriminability decreased. The obtained pattern of bias was directly related to the obtained reinforcer ratio. This research indicates that criterion location measures seem to be preferable indices of response bias.     

Concurrent variation of response bias and sensitivity in an operant-psychophysical test

The yes-no signal detection procedure was applied to a single-response operant paradigm in which rats discriminated between a standard auditory intensity and attenuated comparison values. The payoff matrix was symmetrical (with reinforcing brain stimulation for correct detections and brief time-out for errors), but signal probability and intensity differences were varied to generate a family of isobias and isosensitivity functions. The d’ parameter remained fairly constant across a wide range of bias levels. Isobias functions deviated from a strict matching strategy as discrimination difficulty increased, although an orderly relation was maintained between signal probability value and the degree and direction of response bias.     

On the handling of heavy bias in a self-paced task

Following an earlier observation that systematic variations in performance on a task with heavy frequency imbalance was correlated with fluctuations in the amount of bias in the input programme, an experiment was carried out to establish the roles of long and short term sampling. Two groups of subjects were trained on a self-paced, five-choice task. One group's input had an average of 68 per cent. bias on one source, the other had 44 per cent. on the same source. Analysis of data was carried out on three levels of local bias for each condition, one level being identical for both conditions. It was found that responses to the biased stimuli were determined by the average bias in each input sequence and not by moment to moment variations in that bias. The effect observed originally can therefore be accounted for in terms of a relatively simple additive model which includes the “repetition effect” first described by Bertelson.     

Belief bias is stronger when reasoning is more difficult

Three studies examine the influence of varying the difficulty of reasoning on the extent of belief bias, while minimising the possibility that the manipulation would influence the way participants approach the task. Specifically, reasoning difficulty was manipulated by making variations in problem content, while maintaining all other aspects of the problems constant. In Study 1, 191 participants were presented with consistent and conflict problems varying in two levels of difficulty. The results showed a significant influence of problem difficulty on the extent of the belief bias, such that the effect of belief was more pronounced for difficult problems. This effect was stronger in Study 2 (73 participants) where the difference in the difficulty of the problems was purposely accentuated. The results of both studies stress the importance of controlling for problem difficulty when studying belief bias. Study 3 examined one consequence of this, i.e., the classic belief vs. logic interaction could be eliminated by manipulating problem difficulty. Theoretical implications for dual-process accounts of belief bias are also discussed.     

Isobias curves in some detection tasks

Signal/noise ratio was varied in three tone-detection tasks while other conditions were held constant, generating 9 8-point and 12 12-point isobias plots. These plots were compared with the theoretical isobias curves associated with seven proposed bias measures. Each of the proposed measures covaried with signal/noise ratio in at least half of the data. For two of the plots, none of the measures was invariant. It is clear that different subjects respond quite differently to change in discriminability, and the implications of this for bias measurement are discussed.     

Combining versus analyzing multiple causes: how domain assumptions and task context affect integration rules

In everyday life, people typically observe fragments of causal networks. From this knowledge, people infer how novel combinations of causes they may never have observed together might behave. I report on 4 experiments that address the question of how people intuitively integrate multiple causes to predict a continuously varying effect. Most theories of causal induction in psychology and statistics assume a bias toward linearity and additivity. In contrast, these experiments show that people are sensitive to cues biasing various integration rules. Causes that refer to intensive quantities (e.g., taste) or to preferences (e.g., liking) bias people toward averaging the causal influences, whereas extensive quantities (e.g., strength of a drug) lead to a tendency to add. However, the knowledge underlying these processes is fallible and unstable. Therefore, people are easily influenced by additional task-related context factors. These additional factors include the way data are presented, the difficulty of the inference task, and transfer from previous tasks. The results of the experiments provide evidence for causal model and related theories, which postulate that domain-general representations of causal knowledge are influenced by abstract domain knowledge, data-driven task factors, and processing difficulty.     

Treatments of bias in detection and recognition models: A review

The review considers various proposals about the form of isobias curves and about the relation between bias and bias conditions (payoff, presentation probability, and instructions). Though the proposals differ sharply, none of them proves to be adequately supported by existing evidence. There has been a curious tendency to appeal to a priori arguments, to consider only a very restricted set of alternatives from among the published proposals, to contrast proposals at levels of performance where their predictions are minimally different, and to ignore existing evidence. The review tries to provide a solid basis for more adequate experimental work in the future.     

Residual sensory capacities of the deaf: a signal detection analysis of a visual discrimination task.

This experiment compared the visual sensory sensitivity of deaf and hearing subjects in a signal detection paradigm. Subjects (ns = 6) were required to give forced-choice responses to a brightness discrimination task under three stimulus probability conditions (0.25, 0.50, and 0.75). A total of 1,800 trials were given to each subject and utilized to construct isosensitivity functions and d' and Beta, indices for sensory sensitivity and response bias, respectively. The results showed that no enhanced sensory sensitivity is present for these deaf children and questions the classical sensory compensation hypothesis. Furthermore, the deaf subjects responded in a relatively bias-free manner to variations in stimulus probability.     

Effect of subliminal visual material on an auditory signal detection task

An experiment assessed the effect of subliminally embedded, visual material on an auditory detection task. 22 women and 19 men were presented tachistoscopically with words designated as "emotional" or "neutral" on the basis of prior GSRs and a Word Rating List under four conditions: (a) Unembedded Neutral, (b) Embedded Neutral, (c) Unembedded Emotional, and (d) Embedded Emotional. On each trial subjects made forced choices concerning the presence or absence of an auditory tone (1000 Hz) at threshold level; hits and false alarm rates were used to compute non-parametric indices for sensitivity (A') and response bias (B"). While over-all analyses of variance yielded no significant differences, further examination of the data suggests the presence of subliminally "receptive" and "non-receptive" subpopulations.     

fMRI reliability: Influences of task and experimental design

As scientists, it is imperative that we understand not only the power of our research tools to yield results, but also their ability to obtain similar results over time. This study is an investigation into how common decisions made during the design and analysis of a functional magnetic resonance imaging (fMRI) study can influence the reliability of the statistical results. To that end, we gathered back-to-back test–retest fMRI data during an experiment involving multiple cognitive tasks (episodic recognition and two-back working memory) and multiple fMRI experimental designs (block, event-related genetic sequence, and event-related m-sequence). Using these data, we were able to investigate the relative influences of task, design, statistical contrast (task vs. rest, target vs. nontarget), and statistical thresholding (unthresholded, thresholded) on fMRI reliability, as measured by the intraclass correlation (ICC) coefficient. We also utilized data from a second study to investigate test–retest reliability after an extended, six-month interval. We found that all of the factors above were statistically significant, but that they had varying levels of influence on the observed ICC values. We also found that these factors could interact, increasing or decreasing the relative reliability of certain Task × Design combinations. The results suggest that fMRI reliability is a complex construct whose value may be increased or decreased by specific combinations of factors.     

On the relation between transfer and difficulty of initial task

Subjects were required to throw at a target under three conditions of varying difficulty. Six groups of 14 subjects were used, each group receiving one of the six possible orders of presentation of the three conditions.

Results in terms of mean distance from the “bull” indicated that the three conditions were producing different transfer effects. These could not be adequately explained in terms of stimulus-or response-similarity, and accordingly a tentative hypothesis is put forward in terms of the relative difficulty of the condition presented first, to those presented subsequently. It is suggested that transfer tends to be positive from a relatively difficult initial task to a subsequent task which is easier, while transfer will tend to be negative from a relatively easy initial task to one which is more difficult.

The first of the three conditions presented exerted considerably more powerful transfer effects than the second. Three possibilities as to why this should be so, are outlined.     

A self-fulfilling prophecy in a competitive psychomotor task

Fifty-six pairs of male college students competed in games of electronic table tennis. The participants were led to expect a bias in the game equipment, such that one player would have an advantage and the other a disadvantage. All participants recorded levels of aspiration both before and after being informed of the alleged bias. Upon completion of the game, the subjects made ratings of causal attribution to five factors: ability, task difficulty, effort, luck, and equipment bias. The results showed that the students with the putative disadvantage set significantly lower aspiration levels and performed at a significantly lower level than those students with the supposed advantage. Thus the putative bias became a real bias, although a psychological rather than a physical one. It is suggested that the source of the bias and the performance differences was a self-fulfilling prophecy. In contrast to the results of earlier studies, there were no meaningful differences in ratings of causal attribution.     

The manipulation of color response times in a color-word interference task

The relationship between word and color response times and color-word interference is of theoretical importance for understanding the interference phenomenon. In this experiment, color response times were manipulated by varying the size of the color-word set, but there was no corresponding change in interference. Since the manipulation of word response times consistently leads to such changes, this result has certain implications which were briefly discussed.     

A dual task response modality effect: Support for multiprocessor models of attention

In Experiment I two groups of 11 men performed a continuous visual input/manual output task simultaneously with a two-choice tone identification task. One group responded vocally to the tones; one group responded with the hand not involved in the continuous tracking task. In either perceptual or stimulus uncertainty terms the two combinations were identical; the only difference between them was the modality of the two-choice responses. The continuous task was performed significantly worse when the two-choice responses were manual. The probability of response production on the continuous task was affected by the production of manual responses but not by the production of vocal responses. It was concluded that although the two manual responses were produced by a single limited capacity process, the manual and vocal responses were produced by independent processes. In Experiment II the same manual tracking task was combined with a mental arithmetic task at two levels of difficulty. Tracking performance was independent of the difficulty of the arithmetic task. These results support a multi-processor approach to attention as opposed to single channel models. Results of dual task studies which have used only one pair of response modalities are re-examined in the light of the response modality effect found in Experiment I.     

Strategies of rule discovery in an inference task

It has long been known that subjects in certain inference tasks will seek evidence which can confirm their present hypotheses, even in situations where disconfirmatory evidence could be more informative. We sought to alter this tendency in a series of experiments which employed a rule discovery task, the 2-4-6 problem first described by Wason. The first experiment instructionally modified subjects confirmatory tendencies. While a disconfirmatory strategy was easily induced, it did not lead to greater efficiency in discovering the rule. The second experiment introduced subjects to the possibility of disconfirmation only after they had developed a strongly held hypothesis through the use of confirmatory evidence. This manipulation also failed to alter the efficiency of rule discovery. In the third experiment, subjects were taught to use multiple hypotheses at each step, in the manner of Platt's “Strong Inference”. This operation actually worsened performance. Finally, in the fourth experiment, the structure of the problem was altered slightly by asking subjects to seek two interrelated rules. A dramatic increase in performance resulted, perhaps because information which in previous tasks was seen as merely erroneous could now be related to an alternative rule. The four studies have broad implications for the psychological study of inference processes in general, and for the study of scientific inference in particular.     

Fitting the psychometric function

A constrained generalized maximum likelihood routine for fitting psychometric functions is proposed, which determines optimum values for the complete parameter set--that is, threshold and slope--as well as for guessing and lapsing probability. The constraints are realized by Bayesian prior distributions for each of these parameters. The fit itself results from maximizing the posterior distribution of the parameter values by a multidimensional simplex method. We present results from extensive Monte Carlo simulations by which we can approximate bias and variability of the estimated parameters of simulated psychometric functions. Furthermore, we have tested the routine with data gathered in real sessions of psychophysical experimenting.