Prediction and uncertainty in associative learning: examining controlled and automatic components of learned attentional biases

It has been suggested that attention is guided by two factors that operate during associative learning: a predictiveness principle, by which attention is allocated to the best predictors of outcomes, and an uncertainty principle, by which attention is allocated to learn about the less known features of the environment. Recent studies have shown that predictiveness-driven attention can operate rapidly and in an automatic way to exploit known relationships. The corresponding characteristics of uncertainty-driven attention, on the other hand, remain unexplored. In two experiments we examined whether both predictiveness and uncertainty modulate attentional processing in an adaptation of the dot probe task. This task provides a measure of automatic orientation to cues during associative learning. The stimulus onset asynchrony of the probe display was manipulated in order to explore temporal characteristics of predictiveness- and uncertainty-driven attentional effects. Results showed that the predictive status of cues determined selective attention, with faster attentional capture to predictive than to non-predictive cues. In contrast, the level of uncertainty slowed down responses to the probe regardless of the predictive status of the cues. Both predictiveness- and uncertainty-driven attentional effects were very rapid (at 250?ms from cue onset) and were automatically activated.     

Age differences in associative and strategic processes in human conditional learning

Research indicates associative and strategic deficits mediate age related deficits in memory, whereas simple associative processes are independent of strategic processing and strategic processes mediate resistance to interference. The present study showed age-related deficits in a contingency learning task, although older participants' resistance to interference was not disproportionately affected. Recognition memory predicted discrimination, whereas general cognitive ability predicted resistance to interference, suggesting differentiation between associative and strategic processes in learning and memory, and age declines in associative processes. Older participants' generalisation of associative strength from existing to novel stimulus–response associations was consistent with elemental learning theories, whereas configural models predicted younger participants' responses. This is consistent with associative deficits and reliance on item-level representations in memory during later life.     

Overt attention and predictiveness in human contingency learning

Two experiments used eye-tracking procedures to investigate the relationship between attention and associative learning in human participants. These experiments found greater overt attention to cues experienced as predictive of the outcomes with which they were paired, than to cues experienced as nonpredictive. Moreover, this attentional bias persisted into a second training phase when all cues were equally predictive of the outcomes with which they were paired, and it was accompanied by a related bias in the rate of learning about these cues. These findings are consistent with the attentional model of associative learning proposed by Mackintosh (1975), but not with that proposed by Pearce and Hall (1980).     

Trait mindfulness and attention to emotional information: An eye tracking study

Biased attention for emotional information is associated with the emotional disorders. Trait mindfulness is associated with lower depression and anxiety and with improved attentional control. Mindfulness is also related to lower levels of brooding rumination. The current study examined the association between trait mindfulness, brooding rumination, depressed and anxious state moods, and attention to emotional visual stimuli utilizing eye tracking methodology. Participants were 158 undergraduates. Trait mindfulness was negatively associated with attention to sad and threatening stimuli, but was not associated with attention to positive or neutral stimuli. There was an indirect effect of mindfulness on attention to sad stimuli through brooding rumination. Data are cross sectional but provide initial evidence that mindfulness may partially exert its effects on depression and anxiety by lessening attention to negatively-valenced stimuli.     

Probability judgment in hierarchical learning: a conflict between predictiveness and coherence

Why are people's judgments incoherent under probability formats? Research in an associative learning paradigm suggests that after structured learning participants give judgments based on predictiveness rather than normative probability. This is because people's learning mechanisms attune to statistical contingencies in the environment, and they use these learned associations as a basis for subsequent probability judgments. We introduced a hierarchical structure into a simulated medical diagnosis task, setting up a conflict between predictiveness and coherence. Thus, a target symptom was more predictive of a subordinate disease than of its superordinate category, even though the latter included the former. Under a probability format participants tended to violate coherence and make ratings in line with predictiveness; under a frequency format they were more normative. These results are difficult to explain within a unitary model of inference, whether associative or frequency-based. In the light of this, and other findings in the judgment and learning literature, a dual-component model is proposed.     

Learned predictiveness training modulates biases towards using boundary or landmark cues during navigation

A number of navigational theories state that learning about landmark information should not interfere with learning about shape information provided by the boundary walls of an environment. A common test of such theories has been to assess whether landmark information will overshadow, or restrict, learning about shape information. Whilst a number of studies have shown that landmarks are not able to overshadow learning about shape information, some have shown that landmarks can, in fact, overshadow learning about shape information. Given the continued importance of theories that grant the shape information that is provided by the boundary of an environment a special status during learning, the experiments presented here were designed to assess whether the relative salience of shape and landmark information could account for the discrepant results of overshadowing studies. In Experiment 1, participants were first trained that either the landmarks within an arena (landmark-relevant), or the shape information provided by the boundary walls of an arena (shape-relevant), were relevant to finding a hidden goal. In a subsequent stage, when novel landmark and shape information were made relevant to finding the hidden goal, landmarks dominated behaviour for those given landmark-relevant training, whereas shape information dominated behaviour for those given shape-relevant training. Experiment 2, which was conducted without prior relevance training, revealed that the landmark cues, unconditionally, dominated behaviour in our task. The results of the present experiments, and the conflicting results from previous overshadowing experiments, are explained in terms of associative models that incorporate an attention variant.     

Context-outcome associations mediate context-switch effects in a human predictive learning task

Four experiments explored the role of contexts in information retrieval after different levels of acquisition training in human predictive learning. Participants were trained where cue (X) was followed by an outcome in context A while a different cue (Y) was followed by the absence of the outcome in context B. When 4 training trials with each cue were conducted, testing the stimuli in the alternative contexts decreased predictive judgments to X and increased predictive judgments to Y. These effects disappeared both when training was increased up to 18 trials (Experiments 1a and 1b), and when the outcome was presented in both contexts A and B (Experiments 2 and 4). When the outcome was presented in both contexts, the nonreinforced cue Y, trained in the presumably excitatory context B, became a conditioned inhibitor (Experiment 3). Additional experience with one of the contexts, but not with both, made the context-switch effect reappear (Experiment 4). These results suggest that irrelevant contexts may enter into direct associations with the outcome before prolonged training leads participants to discard them as predictors.     

Attention and expectation in human predictive learning: The role of uncertainty

Three localized, visual pattern stimuli were trained as predictive signals of auditory outcomes. One signal partially predicted an aversive noise in Experiment 1 and a neutral tone in Experiment 2, whereas the other signals consistently predicted either the occurrence or absence of the noise. The expectation of the noise was measured during each signal presentation, and only participants for whom this expectation demonstrated contingency knowledge showed differential attention to the signals. Importantly, when attention was measured by visual fixations, the contingency-aware group attended more to the partially predictive signal than to the consistent predictors in both experiments. This profile of visual attention supports the Pearce and Hall (1980) Pearce, J. M. and Hall, G. 1980. A model for Pavlovian learning: Variations in the effectiveness of conditioned but not unconditioned stimuli. Psychological Review, 87: 532–552. [Crossref], [PubMed], [Web of Science ®] , [Google Scholar] theory of the role of attention in associative learning.     

Using variability to guide dimensional weighting: associative mechanisms in early word learning

At 14 months, children appear to struggle to apply their fairly well-developed speech perception abilities to learning similar sounding words (e.g., bih/dih; Stager & Werker, 1997). However, variability in nonphonetic aspects of the training stimuli seems to aid word learning at this age. Extant theories of early word learning cannot account for this benefit of variability. We offer a simple explanation for this range of effects based on associative learning. Simulations suggest that if infants encode both noncontrastive information (e.g., cues to speaker voice) and meaningful linguistic cues (e.g., place of articulation or voicing), then associative learning mechanisms predict these variability effects in early word learning. Crucially, this means that despite the importance of task variables in predicting performance, this body of work shows that phonological categories are still developing at this age, and that the structure of noninformative cues has critical influences on word learning abilities.     

Effects of moderate chronic ethanol consumption on hippocampal nicotinic receptors and associative learning

A number of studies have reported that ethanol exposure induces changes in different brain systems. The hippocampus is a brain region that is very vulnerable to ethanol exposition, which functionally results in impairment of learning and memory processes reported in heavy drinkers. Hippocampal nicotinic receptors are involved in learning and memory. In this study, we determined the effects of ethanol on the main hippocampal subtypes of neural nicotinic receptors (α7 and α4β2) in rats non-selected for alcohol consumption, in order to check for possible changes on these receptors that could be linked with alterations in learning acquisition. Binding assays were carried out with [³H]methyllycaconitine ([³H]MLA) to study the α7 and [³H]nicotine to study α4β2 receptors. Auto-shaping, continuous ratio and extinction procedures were used as behavioral tests. The results show that moderate chronic ethanol consumption for 10 weeks produces: (a) a decrease of both hippocampal nicotinic receptor subtypes without alterations in affinity; (b) no differences in behavioral performance between control rats and ethanol-drinking rats in auto-shaping and continuous ratio; (c) an improvement of performance of extinction paradigm. These results indicate that chronic ethanol consumption, at moderate levels, induces changes in hippocampal nicotinic receptors but does not impair acquisition and performance of new associative learning and even improves some kind of paradigms. These results may have implications in the biochemical basis of interactions between alcohol and nicotine and the effects of these drugs on behavior.     

Predicting performance on the Raven's Matrices: The roles of associative learning and retrieval efficiency

Previous studies have shown that performance on Williams and Pearlberg's (2006) complex associative learning task is a good predictor of fluid intelligence. This task is similar in structure to that used in studying the fan effect (Anderson, 1974), as both tasks involve forming multiple associations and require retrieval in the face of interference. The purpose of the present study was to investigate the relations among complex associative learning, working memory, and fluid intelligence. Specifically, we asked whether retrieval efficiency, as measured by the fan effect, could account for the relation between complex associative learning and performance on Raven's Advanced Progressive Matrices. Consistent with previous findings, complex associative learning predicted Raven's performance, but the fan effect did not account for this relation. Notably, the learning phase of the fan effect task was significantly correlated with both complex associative learning and Raven's performance, providing further support for the importance of learning as a predictor of fluid intelligence.     

Sense of agency, associative learning, and schizotypy

Despite the fact that the role of learning is recognised in empirical and theoretical work on sense of agency (SoA), the nature of this learning has, rather surprisingly, received little attention. In the present study we consider the contribution of associative mechanisms to SoA. SoA can be measured quantitatively as a temporal linkage between voluntary actions and their external effects. Using an outcome blocking procedure, it was shown that training action–outcome associations under conditions of increased surprise augmented this temporal linkage. Moreover, these effects of surprise were correlated with schizotypy scores, suggesting that individual differences in higher level experiences are related to associative learning and to its impact on SoA. These results are discussed in terms of models of SoA, and our understanding of disrupted SoA in certain disorders.     

Attentional biases in dysphoria: An eye-tracking study of the allocation and disengagement of attention

This study looked for evidence of biases in the allocation and disengagement of attention in dysphoric individuals. Participants studied images for a recognition memory test while their eye fixations were tracked and recorded. Four image types were presented (depression-related, anxiety-related, positive, neutral) in each of two study conditions. For the simultaneous study condition, four images (one of each type) were presented simultaneously for 10 seconds, and the number of fixations and the total fixation time to each image was measured, similar to the procedure used by Eizenman et al. (2003) and Kellough, Beevers, Ellis, and Wells (2008). For the sequential study condition, four images (one of each type) were presented consecutively, each for 4 seconds; to measure disengagement of attention an endogenous cuing procedure was used (Posner, 1980). Dysphoric individuals spent significantly less time attending to positive images than non-dysphoric individuals, but there were no group differences in attention to depression-related images. There was also no evidence of a dysphoria-related bias in initial shifts of attention. With respect to the disengagement of attention, dysphoric individuals were slower to disengage their attention from depression-related images. The recognition memory data showed that dysphoric individuals had poorer memory for emotional images, but there was no evidence of a conventional mood-congruent memory bias. Differences in the attentional and memory biases observed in depressed and dysphoric individuals are discussed.     

Selective attention in human associative learning and recognition memory

Four experiments examined the role of selective attention in a new causal judgment task that allowed measurement of both causal strength and cue recognition. In Experiments 1 and 2, blocking was observed; pretraining with 1 cue (A) resulted in reduced learning about a 2nd cue (B) when those 2 cues were trained in compound (AB+). Participants also demonstrated decreased recognition performance for the causally redundant Cue B, suggesting that less attention had been paid to it in training. This is consistent with the idea that attention is preferentially allocated toward the more predictive Cue A, and away from the less predictive Cue B (e.g., N. J. Mackintosh, 1975). Contrary to this hypothesis, in Experiments 3 and 4, participants demonstrated poorer recognition for the most predictive cues, relative to control cues. A new model, which is based on N. J. Mackintosh's (1975) model, is proposed to account for the observed relationship between the extent to which each cue is attended to, learned about, and later recognized     

Early detection and avoidance of threatening faces during passive viewing

To evaluate whether there is an early attentional bias towards negative stimuli, we tracked participants' eyes while they passively viewed displays composed of four Ekman faces. In Experiment 1 each display consisted of three neutral faces and one face depicting fear or happiness. In half of the trials, all faces were inverted. Although the passive viewing task should have been very sensitive to attentional biases, we found no evidence that overt attention was biased towards fearful faces. Instead, people tended to actively avoid looking at the fearful face. This avoidance was evident very early in scene viewing, suggesting that the threat associated with the faces was evaluated rapidly. Experiment 2 replicated this effect and extended it to angry faces. In sum, our data suggest that negative facial expressions are rapidly analysed and influence visual scanning, but, rather than attract attention, such faces are actively avoided.     

Training attention: Interactions between central cues and reflexive attention

Three experiments are reported in which we investigate whether the recently reported interactions between central cues (e.g., arrows) and reflexive attention are attributable to the overlearned spatial properties of certain central cues. In all three experiments, a nonpredictive cue with arbitrary spatial properties (a colour patch) is presented prior to a detection target in the left or right visual field. Reaction times to detect targets are compared before and after a training session in which participants are trained to associate each colour patch with left and right space, either via a target detection task in which colour predicts target location 100% of the time (Experiments 1 and 3), or via a left/right motor movement as a function of colour (Experiments 2 and 3). In the first two experiments, a small but highly significant training effect is observed. Participants are approximately 10 ms faster to detect targets at congruent locations relative to incongruent locations after training relative to before training, despite the fact that cue colour was nonpredictive during the test sessions. In Experiment 3, the length of the training session is increased and the magnitude of the training effect also increases as a result. Implications for the interaction between central cues and reflexive attention, as well as premotor theory of attention, are discussed.     

Hemispheric asymmetry and human associative learning: Interactions with attention

In a previous study (Hugdahl & Brobeck, 1986) it was shown that Pavlovian conditioning to an auditory verbal conditioned stimulus (CS) initially presented only to the left cerebral hemisphere was stronger than when the same CS was presented to the right hemisphere. This was followed up in the present study by controlling for the possibility that the effect was caused by laterally biased attention. The study was performed using the “dichotic extinction paradigm,” which consists of three different phases. During the habituation phase, the CS+ and CS- were presented binaurally and separated in time. During the acquisition phase, the CS+ was followed by a white-noise unconditioned stimulus (UCS). During the dichotic extinction phase, the CS+ and CS- were presented dichotically, i.e., simultaneous presentations on each trial. Half of the subjects had the CS+ in the right ear, and half had the CS+ in the left ear. Each group was further divided into two subgroups, with one subgroup instructed to attend only to the right ear input, and the other subgroup to attend only to the left ear input. During acquisition, larger electrodermal responses were obtained to the CS+ than to the CS-. During dichotic extinction, the CS+ right ear group showed superior resistance to extinction compared to the CS+ left ear group, with no effect of the manipulation of attention. The effect was, however, attenuated when levels of acquisition was used as covariates in an analysis of covariance. There were overall larger responses from the left hand recording.     

Speculations on the emergence of self-awareness in big-brained organisms: the roles of associative memory and learning, existential and religious questions, and the emergence of tautologies

This paper argues that self-awareness emerges in organisms whose brains have a sufficiently integrated, complex ability for associative learning and memory. Continual sensory input of information related to the organism leads to the formation of a set of associations that may be termed an organismal “self-image”. After providing the basic mechanistic basis for the emergence of an organismal self-image, this paper proceeds to go through a representative list of behaviors associated with self-awareness, and shows how associative memory and learning, combined with an organismal self-image, leads to the emergence of these various behaviors. This paper also discusses various tautologies that invariably emerge when discussing self-awareness. We continue with various speculations on manipulating self-awareness, and discuss how concepts from set and logic theory may provide a useful set of tools for understanding the emergence of higher cognitive functions in complex organisms.     

Errors, efficiency, and the interplay between attention and category learning

Learning to identify objects as members of categories is an essential cognitive skill and learning to deploy attention effectively is a core component of that process. The present study investigated an assumption imbedded in formal models of categorization: error is necessary for attentional learning. Eye-trackers were used to record participants’ allocation of attention to task relevant and irrelevant features while learning a complex categorization task. It was found that participants optimized their fixation patterns in the absence of both performance errors and corrective external feedback. Optimization began immediately after each category was mastered and continued for many trials. These results demonstrate that error is neither necessary nor sufficient for all forms of attentional learning.     

Associative isolation: Unifying associative and list memory

Rather than treating paired associate and serial learning as involving the acquisition of distinct types of information [e.g. Murdock (1974). Human memory: Theory and data. New York: Wiley] I propose an Isolation Principle which treats the two as ends of a continuum. According to this principle, consecutive pairs of items are relatively isolated from other studied items in paired associates learning, but not isolated in serial list learning. The consequence is that variability that dominates forward and backward probed recall is highly correlated in pairs but less so, due to differential interference, in lists. This can explain an important dissociation: whereas forward and backward probes of pairs are nearly perfectly correlated, the correlation is only moderate for serial lists. I demonstrate this in a chaining model by varying item-to-item associative strengths and in a positional coding model by varying the similarity structure of item positions. This enables a range of models to account for data on pairs and lists, as well as potential intermediate or hybrid paradigms, within a single theoretical framework.