Forward blocking in human learning sometimes reflects the failure to encode a cue–outcome relationship

Forward and backward blocking in statistical learning

Prediction errors have a prominent role in many forms of learning. For example, in reinforcement learning, agents learn by updating the association between states and outcomes as a function of the prediction error elicited by the event. One paradigm often used to study error-driven learning is blocking. In forward blocking, participants are first presented with stimulus A, followed by outcome X (A→X). In the second phase, A and B are presented together, followed by X (AB→X). Here, A→X blocks the formation of B→X, given that X is already fully predicted by A. In backward blocking, the order of phases is reversed. Here, the association between B and X that is formed during the first learning phase of AB→X is weakened when participants learn exclusively A→X in the second phase. The present study asked the question whether forward and backward blocking occur during visual statistical learning, i.e., the incidental learning of the statistical structure of the environment. In a series of studies, using both forward and backward blocking, we observed statistical learning of temporal associations among pairs of images. While we found no forward blocking, we observed backward blocking, thereby suggesting a retrospective revaluation process in statistical learning and supporting a functional similarity between statistical learning and reinforcement learning.

Hearing hooves, thinking zebras: A review of the inverse base-rate effect

People often fail to use base-rate information appropriately in decision-making. This is evident in the inverse base-rate effect, a phenomenon in which people tend to predict a rare outcome for a new and ambiguous combination of cues. While the effect was first reported in 1988, it has recently seen a renewed interest from researchers concerned with learning, attention and decision-making. However, some researchers have raised concerns that the effect arises in specific circumstances and is unlikely to provide insight into general learning and decision-making processes. In this review, we critically evaluate the evidence for and against the main explanations that have been proposed to explain the effect, and identify where this evidence is currently weak. We argue that concerns about the effect are not well supported by the data. Instead, the evidence supports the conclusion that the effect is a result of general mechanisms that provides a useful opportunity to understand the processes involved in learning and decision making. We discuss gaps in our knowledge and some promising avenues for future research, including the relevance of the effect to models of attentional change in learning, an area where the phenomenon promises to contribute new insights.

Can We Set Aside Previous Experience in a Familiar Causal Scenario?

Causal and predictive learning research often employs intuitive and familiar hypothetical scenarios to facilitate learning novel relationships. The allergist task, in which participants are asked to diagnose the allergies of a fictitious patient, is one example of this. In such studies, it is common practice to ask participants to ignore their existing knowledge of the scenario and make judgments based only on the relationships presented within the experiment. Causal judgments appear to be sensitive to instructions that modify assumptions about the scenario. However, the extent to which prior knowledge continues to affect competition for associative learning, even after participants are instructed to disregard it, is unknown. To answer this, we created a cue competition design that capitalized on prevailing beliefs about the allergenic properties of various foods. High and low allergenic foods were paired with foods moderately associated with allergy to create two compounds; high + moderate and low + moderate. We expected high allergenic foods to produce greater competition for associative memory than low allergenic foods. High allergenic foods may affect learning either because they generate a strong memory of allergy or because they are more salient in the context of the task. We therefore also manipulated the consistency of the high allergenic cue-outcome relationship with prior beliefs about the nature of the allergies. A high allergenic food that is paired with an inconsistent allergenic outcome should generate more prediction error and thus more competition for learning, than one that is consistent with prior beliefs. Participants were instructed to either use or ignore their knowledge of food allergies to complete the task. We found that while participants were able to set aside their prior knowledge when making causal judgments about the foods in question, associative memory was weaker for the cues paired with highly allergenic foods than cues paired with low allergenic foods regardless of instructions. The consistency manipulation had little effect on this result, suggesting that the effects in associative memory are most likely driven by selective attention to highly allergenic cues. This has implications for theories of causal learning as well as the way causal learning tasks are designed.

Testing the deductive inferential account of blocking in causal learning

The sensitivity of the blocking effect to outcome additivity pretraining has been used to argue that the phenomenon is the result of deductive inference, and to draw general conclusions about the nature of human causal learning. In two experiments, we manipulated participants' assumptions about the additivity of the outcome using pretraining before a typical blocking procedure. Ratings measuring causal judgments, confidence, and expected severity of the outcome were used concurrently to investigate how pretraining affected assumptions of outcome additivity and blocking. In Experiment 1, additive pretraining led to lower causal ratings and higher confidence ratings of the blocked cue, relative to control cues, consistent with the notion that additive pretraining encourages deductive reasoning. However, Experiments 1 and 2 showed that removing additivity assumptions through nonadditive pretraining had no impact on a statistically reliable blocking effect observed in a blocking procedure with no pretraining. We found no evidence that the blocking effect in the absence of pretraining was related to the participants' assumptions about the additivity of the outcome. Although additive pretraining may enhance blocking by encouraging deductive reasoning about the blocked cue, the evidence suggests that blocking in causal learning is not reliant on this reasoning and that humans do not readily engage in deduction merely because they possess the assumptions that permit its use.

A dissociation between causal judgement and the ease with which a cause is categorized with its effect

The associative view of human causal learning argues that causation is attributed to the extent that the putative cause activates, via an association, a mental representation of the effect. That is, causal learning is a human analogue of animal conditioning. We tested this associative theory using a task in which a fictitious character suffered from two allergic reactions, rash (O1) and headache (O2). In a conditioned inhibition design with each of these two outcomes (A–O1/AX– and B–O2/BY–), participants were trained that one herbal remedy (X) prevented O1 and that the other (Y) prevented O2. These inhibitory properties were revealed in a causal judgement summation test. In a subsequent categorization task, X was most easily categorized with O1, and Y with O2. Thus, the categorization data indicated an excitatory X–O1 and Y–O2 association, the reverse of the inhibitory relationship observed on the causal judgement measure. A second experiment showed that this pattern of excitation and inhibition is dependent on intermixed A–O1 and AX– trials. These results are problematic for the standard application of associative activation theories to causal judgement. We argue instead that the inhibition revealed in the causal judgement task reflects inferential reasoning, which relies, in part, on the ability of the cue in question to excite a representation of the outcome, as revealed in the categorization test.

Associationism and cognition: Human contingency learning at 25

A major topic within human learning, the field of contingency judgement, began to emerge about 25 years ago following publication of an article on depressive realism by Alloy and Abramson (1979). Subsequently, associationism has been the dominant theoretical framework for understanding contingency learning but this has been challenged in recent years by an alternative cognitive or inferential approach. This article outlines the key conceptual differences between these approaches and summarizes some of the main methods that have been employed to distinguish between them.

Cue Competition Effects in Human Causal Learning

Five experiments involving human causal learning were conducted to compare the cue competition effects known as blocking and unovershadowing, in proactive and retroactive instantiations. Experiment 1 demonstrated reliable proactive blocking and unovershadowing but only retroactive unovershadowing. Experiment 2 replicated the same pattern and showed that the retroactive unovershadowing that was observed was interfered with by a secondary memory task that had no demonstrable effect on either proactive unovershadowing or blocking. Experiments 3a, 3b, and 3c demonstrated that retroactive unovershadowing was accompanied by an inflated memory effect not accompanying proactive unovershadowing. The differential pattern of proactive versus retroactive cue competition effects is discussed in relationship to amenable associative and inferential processing possibilities.

Selectivity in associative learning: a cognitive stage framework for blocking and cue competition phenomena

BLOCKING IS THE MOST IMPORTANT PHENOMENON IN THE HISTORY OF ASSOCIATIVE LEARNING THEORY: for over 40 years, blocking has inspired a whole generation of learning models. Blocking is part of a family of effects that are typically termed "cue competition" effects. Common amongst all cue competition effects is that a cue-outcome relation is poorly learned or poorly expressed because the cue is trained in the presence of an alternative predictor or cause of the outcome. We provide an overview of the cognitive processes involved in cue competition effects in humans and propose a stage framework that brings these processes together. The framework contends that the behavioral display of cue competition is cognitively construed following three stages that include (1) an encoding stage, (2) a retention stage, and (3) a performance stage. We argue that the stage framework supports a comprehensive understanding of cue competition effects.

Evidence that a transient but cognitively demanding process underlies forward blocking

Blocking is a learning phenomenon in which prior experience inhibits learning about novel cues. Though the phenomenon itself has been well documented, the details of blocking-related processes still remain contentious. Two experiments investigated whether participants were engaged in demanding cognitive processing during different portions of a standard blocking paradigm. Participants in Experiment 1 engaged in a simple secondary task while completing a standard blocking procedure. Results showed that performance on the secondary task was briefly diminished early in the second phase of the blocking paradigm, when the novel cue is first paired with the pretrained cue. Participants in Experiment 2 performed a difficult cognitive load task during either the early or the late portions of the second phase of blocking. The blocking effect was eliminated when learners were under load early in the second phase, but not when learners were under load late in the second phase. These results suggest that blocking relies on a cognitively demanding process with a distinct time course. Computational simulations illustrate how a model that includes top-down (i.e., cognitively demanding) attentional modulation can reproduce the observed behaviour. This suggests that purely associative processes are not sufficient to explain the observed behaviour. Implications for current accounts of blocking are discussed.

Backward versus Forward Blocking: Evidence for Performance-Based Models of Human Contingency Learning

Two types of theories are usually invoked to account for cue-interaction effects in human-contingency learning, performance-based theories, such as the comparator hypothesis and statistical models, and learning-based theories, such as associative models. Interestingly, the former models predict two important cue-interaction effects, forward and backward blocking, should affect responding in a similar manner, whereas learning-based models predict the effect of forward blocking should be larger than the effect of backward blocking. Previous experiments involved important methodological problems, and results have been contradictory. The present experiment was designed to explore potential asymmetries between forward and backward blocking. Analyses yielded similar effect sizes, thereby favoring the explanation by performance-based models.

Learning: from association to cognition

Since the very earliest experimental investigations of learning, tension has existed between association-based and cognitive theories. Associationism accounts for the phenomena of both conditioning and "higher" forms of learning via concepts such as excitation, inhibition, and reinforcement, whereas cognitive theories assume that learning depends on hypothesis testing, cognitive models, and propositional reasoning. Cognitive theories have received considerable impetus in regard to both human and animal learning from recent research suggesting that the key illustration of cue selection in learning, blocking, often arises from inferential reasoning. At the same time, a dichotomous view that separates noncognitive, unconscious (implicit) learning from cognitive, conscious (explicit) learning has gained favor. This review selectively describes key findings from this research, evaluates evidence for and against associative and cognitive explanatory constructs, and critically examines both the dichotomous view of learning as well as the claim that learning can occur unconsciously.

Straw-men and selective citation are needed to argue that associative-link formation makes no contribution to human learning

Mitchell et al. contend that there is no need to posit a contribution based on the formation of associative links to human learning. In order to sustain this argument, they have ignored evidence which is difficult to explain with propositional accounts; and they have mischaracterised the evidence they do cite by neglecting features of these experiments that contradict a propositional account.

Short Article: Extinction in human learning and memory

In two experiments, participants were given extinction training in a human causal learning task. In both experiments, three critical experimental cues were paired with different outcomes in a first phase of training and were then extinguished in a second phase. Three control cues were given the same treatment in the first phase of training, but were not then presented in the second phase. Participants’ ability to correctly identify the outcome with which each cue had been paired in the first phase was lower for extinguished than for control cues. Causal attributions to the extinguished cues were also lower than those to the control cues, a difference that correlated with outcome memory. These data are consistent with the idea that extinction in causal judgement is due, at least in part, to a failure to remember the cue–outcome relationship encoded in the first phase of training.