Optimal cue aggregation in the absence of criterion knowledge

The study of multi‐cue judgment investigates how decision makers aggregate cues to predict the value of a criterion variable. We consider a multi‐cue judgment task in which decision makers have prior knowledge of inter‐cue relationships but are ignorant of how the cues correlate with the criterion. In this setting, a naive judgment strategy prescribes weighting the cues equally. Although many participants are well described via an equal weighting scheme, we find that a substantial minority of participants make predictions consistent with a weighting scheme based on a low‐dimensional projection of the cue space that optimally takes into account inter‐cue correlations. The use of such a weighting scheme is consistent with minimizing maximal error in prediction when the cue‐criterion relationships are unknown.     

Individual Differences in Categorization Gradience As Predicted by Online Processing of Phonetic Cues During Spoken Word Recognition: Evidence From Eye Movements

Recent studies have documented substantial variability among typical listeners in how gradiently they categorize speech sounds, and this variability in categorization gradience may link to how listeners weight different cues in the incoming signal. The present study tested the relationship between categorization gradience and cue weighting across two sets of English contrasts, each varying orthogonally in two acoustic dimensions. Participants performed a four-alternative forced-choice identification task in a visual world paradigm while their eye movements were monitored. We found that (a) greater categorization gradience derived from behavioral identification responses corresponds to larger secondary cue weights derived from eye movements; (b) the relationship between categorization gradience and secondary cue weighting is observed across cues and contrasts, suggesting that categorization gradience may be a consistent within-individual property in speech perception; and (c) listeners who showed greater categorization gradience tend to adopt a buffered processing strategy, especially when cues arrive asynchronously in time.     

Cue integration with categories: Weighting acoustic cues in speech using unsupervised learning and distributional statistics

During speech perception, listeners make judgments about the phonological category of sounds by taking advantage of multiple acoustic cues for each phonological contrast. Perceptual experiments have shown that listeners weight these cues differently. How do listeners weight and combine acoustic cues to arrive at an overall estimate of the category for a speech sound? Here, we present several simulations using a mixture of Gaussians models that learn cue weights and combine cues on the basis of their distributional statistics. We show that a cue‐weighting metric in which cues receive weight as a function of their reliability at distinguishing phonological categories provides a good fit to the perceptual data obtained from human listeners, but only when these weights emerge through the dynamics of learning. These results suggest that cue weights can be readily extracted from the speech signal through unsupervised learning processes.     

Reconciling Compensatory and Noncompensatory Strategies of Cue Weighting: A Causal Model Approach

When forming a judgment about any unknown item, people must draw inferences from information that is already known. This paper examines causal relationships between cues as a relevant factor influencing how people determine the amount of weight to place on each piece of available evidence. We propose that people draw from their beliefs about specific causal relationships between cues when determining how much weight to place on those cues, and that understanding this process can help reconcile differences between predictions of compensatory and lexicographic heuristic strategies. As causal relationships change, different cues become more or less important. Across three experiments, we find support for the use of causal models in determining cue weights, but leave open the possibility that they work in concert with other strategies as well. We conclude by discussing relative strengths and weaknesses of the causal model approach relative to existing models, and suggest areas for future research. Copyright © 2016 John Wiley & Sons, Ltd.     

Verbalization of Decision Strategies in Multiple‐Cue Probabilistic Inference

In multiple‐cue probabilistic inference, people choose between alternatives based on several cues, each of which is differentially associated with an alternative's overall value. Various strategies have been proposed for probabilistic inference (e.g., weighted additive, tally, and take‐the‐best). These strategies differ in how many cue values they require to enact and in how they weight each cue. Do decision makers actually use any of these strategies? Ways to investigate this question include analyzing people's choices and the cues that they reveal. However, different strategies often predict the same decisions, and search behavior says nothing about whether or how people use the information that they acquire. In this research, we attempt to elucidate which strategies participants use in a multiple‐cue probabilistic inference task by examining verbal protocols, a high‐density source of process data. The promise of verbal data is in their utility for testing detailed information processing models. To that end, we apply protocol analysis in conjunction with computational simulations. We find converging evidence across outcome measures, search measures, and verbal reports that most participants use simplifying heuristics, namely take‐the‐best. Copyright © 2015 John Wiley & Sons, Ltd.     

The impact of information distribution,ownership, and discussion on group member judgment: The differential cue weighting model

It is often assumed that groups or teams pool knowledge and consider more information than individuals. However, numerous group decision-making studies have found that groups tend to focus on information that members had in common before group discussion instead of exchanging members’ unique information. This paper extends previous research and introduces the differential cue weighting (DCW) model to explain how group members form and revise their judgments during group discussion. It is proposed that information cues are differentially weighted depending on the combination of three factors: the cue’s initial distribution, its ownership by the member, and its addition to group discussion. An empirical study provides support for the DCW model, demonstrating that, for individual judgment, unique information can be just as influential as common information, but only for the member who contributed it to discussion.     

What determines visual cue reliability?

Visual environments contain many cues to properties of an observed scene. To integrate information provided by multiple cues in an efficient manner, observers must assess the degree to which each cue provides reliable versus unreliable information. Two hypotheses are reviewed regarding how observers estimate cue reliabilities, namely that the estimated reliability of a cue is related to the ambiguity of the cue, and that people use correlations among cues to estimate cue reliabilities. Cue reliabilities are shown to be important both for cue combination and for aspects of visual learning.     

Allocation of attention, reading skills, and deafness.

Brannan and Williams (1987) found that poor readers cannot successfully utilize parafoveal cues to identify letter targets. Whether a similar deficit in the use of cue information occurs in deaf poor readers and whether it is only specific to processes that capture attention automatically were investigated in congenitally deaf young adults classified as poor or good readers and hearing controls classified as good readers. Subjects were presented with central or parafoveal cues that varied in cue validity probability, followed by letter targets presented to the left or right of fixation. The reaction time data analyses showed significant main effects for cue type and cue location and significant interactions among cue type, cue location, cue validity probability, and visual field. No significant main effect or interactions involving groups were found. These results raise the possibility that reading difficulties associated with deafness do no involve a deficit in the visual attentional system of deaf people. They also confirm that parafoveal cues are more effective than central cues in capturing attention.     

The owl and the pussycat: Gaze cues and visuospatial orienting

Recent research has shown that nonpredictive gaze cues trigger reflexive shifts in attention toward the looked-at location. But just how generalizable is this spatial cuing effect? In particular, are people especially tuned to gaze cues provided by conspecifics, or can comparable shifts in visual attention be triggered by other cue providers and directional cues? To investigate these issues, we used a standard cuing paradigm to compare the attentional orienting produced by different cue providers (i.e., animate vs. inanimate) and directional cues (i.e., eyes vs. arrows). The results of three experiments revealed that attentional orienting was insensitive to both the identity of the cue provider and the nature of the triggering cue. However, compared with arrows, gaze cues prompted a general enhancement in the efficiency of processing operations. We consider the implications of these findings for accounts of reflexive visual orienting.     

Test of a distributional theory of intuitive numerical prediction

This paper investigates models that describe how subjects combine uncertain information to arrive at an intuitive prediction of a criterion. Subjects were trained, with feedback, to predict a numerical criterion from each of three single cues. Then they were asked to predict the criterion, without feedback, either from pairs of cues or from single cues. Their predictions were not consistent with a relative weight averaging model, since pairs of cues did not combine additively. Instead, the effect of a cue was inversely proportional to the standard deviation of the criterion at each level of the cue. Subjects appeared to apply greater weight to cue levels with smaller variance, i.e., those cue levels that were more valid. The data could be described by a distributional theory referred to as the equal probability model. For the present experiment, this model implies that the criterion means associated with the levels of each cue are weighted by the reciprocals of the standard deviations and then averaged. Relations between the equal probability model and other models of impression formation are discussed.     

Does causal knowledge help us be faster and more frugal in our decisions?

One challenge that has to be addressed by the fast and frugal heuristics program is how people manage to select, from the abundance of cues that exist in the environment, those to rely on when making decisions. We hypothesize that causal knowledge helps people target particular cues and estimate their validities. This hypothesis was tested in three experiments. Results show that when causal information about some cues was available (Experiment 1), participants preferred to search for these cues first and to base their decisions on them. When allowed to learn cue validities in addition to causal information (Experiment 2), participants also became more frugal (i.e., they searched fewer of the available cues), made more accurate decisions, and were more precise in estimating cue validities than was a control group that did not receive causal information. These results can be attributed to the causal relation between the cues and the criterion, rather than to greater saliency of the causal cues (Experiment 3). Overall, our results support the hypothesis that causal knowledge aids in the learning of cue validities and is treated as a meta-cue for identifying highly valid cues.     

Do as I say - or as I do?! How 18- and 24-month-olds integrate words and actions to infer intentions in situations of match or mismatch

Infants use behavioral and verbal cues to infer another person’s action intention. However, it is still unclear how infants integrate these often co-occurring cues depending on the cues’ coherence (i.e., the degree to which the cues provide coherent information about another’s intention). This study investigated how 18- and 24-month-olds’ (N = 88 per age group) action selection was influenced by varying the coherence of a model’s verbal and behavioral cues. Using a between-subjects design, infants received six trials with different stimulus objects. In the conditions Congruent, Incongruent, and Failed-attempt, the model uttered a telic verb particle that was followed by a matching or contradicting goal-directed action demonstration, or by a non goal-directed slipping motion, respectively. In the condition Pseudo-word, a nonsense word was combined with a goal-directed action demonstration. Infants’ action selection indicated an adherence to the verbal cue in Congruent, Incongruent, and Failed-attempt, and this was stronger in 24- than 18-month-olds. Additionally, in Incongruent and Failed-attempt, patterns of cue integration across the six trials varied in the two age groups. Regarding the behavioral cue, infants in Congruent and Pseudo-word preferentially followed this cue in both age groups, which also suggested a rather unspecific effect of the verbal cue in Congruent. Relatively longer first action-latencies in Incongruent and Failed-attempt implied that these types of coherence elicited higher cognitive demands than in Congruent and Pseudo-word. Results are discussed in light of infants’ flexibility in using social cues, depending on the cue’s coherence and on age-related social-cognitive differences.     

Labeling and family resemblance in the discrimination of polymorphous categories by pigeons

Two experiments examined whether pigeons discriminate polymorphous categories on the basis of a single highly predictive feature or overall similarity. In the first experiment, pigeons were trained to discriminate between categories of photographs of complex real objects. Within these pictures, single features had been manipulated to produce a highly salient texture cue. Either the picture or the texture provided a reliable cue for discrimination during training, but in probe tests, the picture and texture cues were put into conflict. Some pigeons showed a significant tendency to discriminate on the basis of the picture cue (overall similarity or family resemblance), whereas others appeared to rely on the manipulated texture cue. The second experiment used artificial polymorphous categories in which one dimension of the stimulus provided a completely reliable cue to category membership, whereas three other dimensions provided cues that were individually unreliable but collectively provided a completely reliable basis for discrimination. Most pigeons came under the control of the reliable cue rather than the unreliable cues. A minority, however, came under the control of single dimensions from the unreliable set. We conclude that cue salience can be more important than cue reliability in determining what features will control behavior when multiple cues are available.     

Do People Order Cues by Retrieval Fluency when Making Probabilistic Inferences?

In multiple‐cue probabilistic inferences, people infer alternatives' unknown values on decision criteria, using alternatives' attributes as cues. Some inferential strategies, like take‐the‐best, assume that people consider relevant cues sequentially in order of decreasing validity. This assumption has been deemed cognitively implausible by some, who suggest memory retrieval principles to guide cue order. We test whether memory‐based inferences are better described by a model considering cues in order of validity or in order of memory retrieval. In an experiment, we manipulated the frequency with which cues appeared in a learning phase, increasing retrieval fluency of cue values related to the more frequently appearing cue. In a subsequent decision phase, participants made a series of two‐alternative decisions based on the learned cue values. We compared two sequential sampling models, which differed in whether cues are sampled in order of subjective cue validity or in order of retrieval fluency. To model retrieval order of cues in the fluency sampling model, we used the declarative memory theory embedded in the ACT‐R cognitive architecture. Most participants' decisions were best described by the model sampling cues in order of memory retrieval. Only a minority of participants were classified as sampling cues by validity. Our result suggests that retrieval fluency is the primary driver of cue order in inferences from memory, irrespective of the cues' validities. Copyright © 2017 John Wiley & Sons, Ltd.     

Effects of alcohol and nicotine on free recall of relevant cues

It is proposed that alcohol and nicotine will affect free recall of relevant cues and that these effects are mediated by changes in level of arousal. Alcohol (0.0 ml, 0.97 ml, or 2.33 ml per kg of body weight) and nicotine (no cigarettes, or two cigarettes each containing 1.3 mg of nicotine) were administered either individually or jointly to six groups of 15 men, such that different combinations of alcohol and nicotine induced different levels of arousal. Since subjects performed at chance level on recall of word-color, which constituted the irrelevant cue, no conclusions regarding the effect of the drugs on recall of irrelevant cues could be drawn. Under conditions of high arousal, induced by the administration of nicotine or a low dose of alcohol, immediate free recall of 20 adjectives was superior to recall in the low-arousal conditions (a moderate dose of alcohol administered either singly or in conjunction with nicotine) and in the control group. This finding is in line with the prediction that arousal, induced by nicotine or a low alcohol dose enhances attention to high-priority task components.     

Developmental changes in the weighting of prosodic cues

Previous research has shown that the weighting of, or attention to, acoustic cues at the level of the segment changes over the course of development (Nittrouer & Miller, 1997; Nittrouer, Manning & Meyer, 1993). In this paper we examined changes over the course of development in weighting of acoustic cues at the suprasegmental level. Specifically, we tested English-learning 4-month-olds' performance on a clause segmentation task when each of three acoustic cues to clausal units was neutralized and contrasted it with performance on a Baseline condition where no cues were manipulated. Comparison with the reported performance of 6-month-olds on the same task (Seidl, 2007) reveals that 4-month-olds weight prosodic cues to clausal boundaries differently than 6-month-olds, relying more heavily on all three correlates of clausal boundaries (pause, pitch and vowel duration) than 6-month-olds do, who rely primarily on pitch. We interpret this as evidence that 4-month-olds use a holistic processing strategy, while 6-month-olds may already be able to attend separately to isolated cues in the input stream and may, furthermore, be able to exploit a language-specific cue weighting. Thus, in a way similar to that in other cognitive domains, infants begin as holistic auditory scene processors and are only later able to process individual auditory cues.     

Are we aware of our ability to forget? Metacognitive predictions of directed forgetting

It is often important to accurately predict not only what information we will later remember, but also what information we are likely to forget. The present research examined these abilities in the context of item-method directed forgetting, to determine whether people are aware of their strategic control of remembering and forgetting, as well as what cues are used when making metacognitive judgments. Participants studied words, each of which was followed by a cue to remember (R) or forget (F) the word for an upcoming test, and also made predictions of how likely they would be to later recall each word. When asked to recall all of the words, regardless of cue, both recall and predicted recall were fairly sensitive to the R or F instructions, despite some overconfidence. A similar and stronger pattern was found when words were assigned positive or negative point values as cues to remember or forget. These findings suggest that item-based cues to remember or forget information can be successfully utilized when making metacognitive judgments, and that people are fairly aware of the control they have over both remembering and forgetting information.     

The role of landmarks and boundaries in the development of spatial memory

It has been suggested that learning an object's location relative to (1) intramaze landmarks and (2) local boundaries is supported by parallel striatal and hippocampal systems, both of which rely upon input from a third system for orientation. However, little is known about the developmental trajectories of these systems' contributions to spatial learning. The present study tested 5- and 7-year-old children and adults on a water maze-like task in which all three types of cue were available. Participants had to remember the location of an object hidden in a circular bounded environment containing a moveable intramaze landmark and surrounded by distal cues. Children performed less accurately than adults, and showed a different pattern of error. While adults relied most on the stable cue provided by the boundary, children relied on both landmark and boundary cues similarly, suggesting a developmental increase in the weighting given to boundary cues. Further, adults were most accurate in coding angular information (dependent on distal cues), whereas children were most accurate in coding distance, suggesting a developing ability to use distal cues to orient. These results indicate that children as young as 5 years use boundary, intramaze landmark, and distal visual cues in parallel, but that the basic accuracy and relative weighting of these cues changes during subsequent development.     

Cue Combinatorics in Memory Retrieval for Anaphora

Many studies have shown that memory retrieval for real‐time language processing relies on a cue‐based access mechanism, which allows the cues available at the retrieval site to directly access the target representation in memory. An open question is how different types of cues are combined at retrieval to create a single retrieval probe (“cue combinatorics”). This study addresses this question by testing whether retrieval for antecedent‐reflexive dependencies combines cues in a linear (i.e., additive) or nonlinear (i.e., multiplicative) fashion. Results from computational simulations and a reading time experiment show that target items that match all the cues of the reflexive are favored more than target items that mismatch these cues, and that different degrees of mismatches slow reading times in comparable amounts. This profile is consistent with the predictions of a nonlinear cue combination and provides evidence against models in which all cues combine in a linear fashion. A follow‐up set of simulations shows that a nonlinear rule also captures previous demonstrations of interference from nontarget items during retrieval for reflexive licensing. Taken together, these results shed new light on how different types of cues combine at the retrieval site and reveal how the method of cue combination impacts the accessibility of linguistic information in memory.     

Using source cues and familiarity cues to resist imagination inflation

To investigate whether people can resist imagination inflation--the imagination-induced increased confidence that fictitious childhood events really happened--we gave them different types of cues. In a three-stage procedure, participants: (1) rated their confidence that a list of childhood events had happened to them, (2) imagined some of these events, and (3) made confidence ratings a second time. Subjects received either no cues about the source of the imagined event, an additional source cue (perspective), an additional familiarity cue (a plausibility questionnaire), or both cues. Only subjects who had both types of cues resisted imagination inflation. These results suggest that additional cues can sometimes safeguard people from becoming more confident that fictitious events were genuine experiences.