Rich sensorimotor interaction facilitates language learning and is presumed to ground conceptual representations. Yet empirical support for early stages of embodied word learning is currently lacking. Finding evidence that sensorimotor interaction shapes learned linguistic representations would provide crucial support for embodied language theories. We developed a gamified word learning experiment in virtual reality in which participants learned the names of six novel objects by grasping and manipulating objects with either their left or right hand. Participants then completed a word-color match task in which they were tested on the same six words and objects. Participants were faster to respond to stimuli in the match task when the response hand was compatible with the hand used to interact with the named object, an effect we refer to as affordance compatibility. In two follow up experiments, we found that merely observing virtual hands interact with the objects was sufficient to acquire a smaller affordance compatibility effect, and we found that the compatibility effect was driven primarily by responses with a compatible hand and not by responses in a compatible spatial location. Our results support theoretical views of language which ground word representations in sensorimotor experiences, and they suggest promising future routes to explore the sensorimotor foundations of higher cognition through immersive virtual experiments.
This paper examines the degree to which binary instant feedback on computer programming exercises, provided by an Automated Assessment (AA) system, benefits students. It also offers an approach to providing improved feedback. Student behavior in an undergraduate computer science class was studied. Students were assigned exercises requiring the generation of programs that met given specifications. We employed an AA system that evaluated the correctness of student code by executing it on a set of test cases. Students promptly received binary (“Correct”/”Incorrect”) feedback, and they could repeatedly resubmit solutions in response. We found that more than half of the students failed to achieve correct solutions within a reasonable time. A small group of students were also found to have plagiarized solutions. This result led us to investigate ways in which AA systems for programming exercises might provide more rich and detailed feedback. We propose the development of clustering algorithms that group solutions based on how similarly incorrect they are. For the exercises we considered, there were, on average, 64 incorrect submissions, but there were only 8-10 distinct logical errors. This means that, if all incorrect submissions were automatically grouped into 8-10 clusters, a human instructor would only have to produce detailed feedback once for each cluster. That feedback could then be automatically delivered in response to each submission that fell within that cluster. We provide evidence that such an approach would result in substantial labor savings, while providing instant detailed feedback to students.
Persons with autism regularly exhibit executive dysfunction (ED), including problems with deliberate goal-directed behavior, planning, and flexible responding in changing environments. Indeed, this array of deficits is sufficiently prominent to have prompted a theory that executive dysfunction is at the heart of these disorders. A more detailed examination of these behaviors reveals, however, that some aspects of executive function remain developmentaly appropriate. In particular, while people with autism often have difficulty with tasks requiring cognitive flexibility, their fundamental cognitive control capabilities, such as those involved in inhibiting an inappropriate but relatively automatic response, show no significant impairment on many tasks. In this article, an existing computational model of the prefrontal cortex and its role in executive control is shown to explain this dichotomous pattern of behavior by positing abnormalities in the dopamine-based modulation of frontal systems in individuals with autism. This model offers excellent qualitative and quantitative fits to performance on standard tests of cognitive control and cognitive flexibility in this clinical population. By simulating the development of the prefrontal cortex, the computational model also offers a potential explanation for an observed lack of executive dysfunction early in life.
Is conceptual space continuous? The answer to this question depends on how concepts are represented in the brain. Vector space representations, which ground conceptual states in the instantaneous firing rates of neurons, have successfully captured cognitive dynamics in a broad range of domains. There is a growing body of evidence, however, that conceptual information is encoded in spatiotemporal patterns of neural spikes, sometimes called polychronous neuronal groups (PNGs). The use of PNGs to represent conceptual states, rather than employing a continuous vector space, introduces new challenges, including issues of temporally extended representations, meaning through symbol grounding, compositionality, and representational similarity. In this article, we explore how PNGs support discontinuous transitions between concepts. While the continuous dynamics of vector space approaches require such transitions to activate intermediate and blended concepts, PNGs offer the means to change the activation of concepts discretely, introducing a form of conceptual dynamics unavailable to vector space models.
The ability to flexibly, rapidly, and accurately perform novel tasks is a hallmark of human behavior. In our everyday lives we are often faced with arbitrary instructions that we must understand and follow, and we are able to do so with remarkable ease. It has frequently been argued that this ability relies on symbol processing, which depends critically on the ability to represent variables and bind them to arbitrary values. Whereas symbol processing is a fundamental feature of all computer systems, it remains a mystery whether and how this ability is carried out by the brain. Here, we provide an example of how the structure and functioning of the prefrontal cortex/basal ganglia working memory system can support variable binding, through a form of indirection (akin to a pointer in computer science). We show how indirection enables the system to flexibly generalize its behavior substantially beyond its direct experience (i.e., systematicity). We argue that this provides a biologically plausible mechanism that approximates a key component of symbol processing, exhibiting both the flexibility, but also some of the limitations, that are associated with this ability in humans.
Human behavior emerges from a complex dynamic interaction between graded and context-sensitive neural processes, the biomechanics of our bodies, and the vicissitudes of our environments. These coupled processes bear little resemblance to the iterated application of simple symbolic rules. Still, there are circumstances under which our behavior appears to be guided by explicit mental rules. A prototypical case is when succinct verbal instructions are communicated and are promptly followed by another. How does the brain support such rule-guided behavior? How are explicit rules represented in the brain? How are rule representations shaped by experience? What neural processes form the foundation of our ability to systematically represent and apply rules from the vast range of possible rules? This article reviews a line of research that has sought a computational cognitive neuroscience account of rule-guided behavior in terms of the functioning of the prefrontal cortex, the basal ganglia, and related brain systems.
Complexity is widespread in neuronal spike trains and propagation of spike activity, in that variations in measurements of neural activity are irregular, heterogeneous, non-stationary, transient, and scale-free. There are numerous possible reasons for this complexity, and numerous possible consequences for neural and behavioral function. The present review is focused on relationships among neural plasticity, learning, and complex spike dynamics in animal nervous systems, including those of humans. The literature on complex spike dynamics and mechanisms of synaptic plasticity are reviewed for the purpose of considering the roles that each might play for the other. That is, the roles of complex spike dynamics in learning and regulatory functions are considered, as well as the roles of learning and regulatory functions in generating complex spike dynamics. Experimental and computational studies from a range of disciplines and perspectives are discussed, and it is concluded that cognitive science and neuroscience have much to gain from investigating the adaptive aspects of complex spike dynamics for neural and cognitive function.
Cognitive agents physically interacting with the world can best adapt to their task environments if they are able to learn motor skills from experience. Many cognitive architectures have focused on a single learning mechanism to accomplish such adaptation. Behavioral studies with humans, however, have shown that the acquisition of a motor skill generally occurs in two stages. In the initial stage, acquisition is performed via attention-demanding neural processes, producing a high cognitive load. This is followed by more fluent automatic processing, requiring less deliberation. Neuroscientific studies have since identified two relevant interacting neural systems, suggesting that the acquisition of a motor skill involves a transition from heavy depen- dence on a system involving cognitive control to only weak dependence on such a system. This cognitive control system, which includes the prefrontal cortex, is thought to be responsible for acquiring and manipulating declarative representations of skills. This frontal system is seen as modulating processing in a more automatic neural pathway, which develops procedural repre- sentations over time. In this paper, we propose a biologically plausible computational model of motor skill automaticity. This model offers a neurocomputational account of the translation of declarative into procedural knowledge during learning. In support of the model, we review some previously reported experimental results, and we demonstrate, through simulation, how the model provides a parsimonious explanation for these results. The model is seen as exemplifying a novel approach to motor skill learning in artificial agents.
Reversal of synaptic plasticity has been the prevalent theory for extinction of animal conditioning. Phenomena like faster reacquisition after extinction are explained via residual synaptic plasticity in the relevant neural circuits. However, this account cannot explain many recent behavioral findings. This includes phenomena like savings in extinction, reinstatement, spontaneous recovery and renewal. These phenomena point to the possibility that extinction is not a mere reversal of the associations formed during acquisition. It instead involves the superimposition of some separate decremental process that works to inhibit the previously learned responses. We have explored this dual-pathway account using a neurocomputational model of conditioning. In our model, associations related to acquisition and extinction are maintained side by side as a result of the interaction between general neural learning processes and the presence of lateral inhibition between neurons. The model captures most of the relevant behavioral phenomena that prompted the hypothesis of separate acquisition and extinction pathways. It also shows how seemingly complex behavior can emerge out of relatively simple underlying neural mechanisms.
The prefrontal cortex (PFC) plays a central role in flexible cognitive control, including the suppression of habitual responding in favour of situation-appropriate behaviours that can be quite novel. PFC provides a kind of working memory, maintaining the rules, goals, and/or actions that are to control behaviour in the current context. For flexible control, these PFC representations must be sufficiently componential to support systematic generalisation to novel situations. The anatomical structure of PFC can be seen as implementing a componential ‘slot-filler’ structure, with different components encoded over isolated pools of neurons. Previous PFC models have highlighted the importance of a dynamic gating mechanism to selectively update individual ‘slot’ contents. In this article, we present simulation results that suggest that systematic generalisation also requires an ‘output gating’ mechanism that limits the influence of PFC on more posterior brain areas to reflect a small number of representational components at any one time.
Connectionist and dynamical systems approaches explain human thought, language and behavior in terms of the emergent consequences of a large number of simple noncognitive processes. We view the entities that serve as the basis for structured probabilistic approaches as abstractions that are occasionally useful but often misleading: they have no real basis in the actual processes that give rise to linguistic and cognitive abilities or to the development of these abilities. Although structured probabilistic approaches can be useful in determining what would be optimal under certain assumptions, we propose that connectionist, dynamical systems, and related approaches, which focus on explaining the mechanisms that give rise to cognition, will be essential in achieving a full understanding of cognition and development.
PDP++ is a freely available, open source software package designed to support the development, simulation, and analysis of research-grade connectionist models of cognitive processes. It supports most popular parallel distributed processing paradigms and artificial neural network architectures, and it also provides an implementation of the LEABRA computational cognitive neuroscience framework. Models are typically constructed and examined using the PDP++ graphical user interface, but the system may also be extended through the incorporation of user-written C++ code. This article briefly reviews the features of PDP++, focusing on its utility for teaching cognitive modeling concepts and skills to university undergraduate and graduate students. An informal evaluation of the software as a pedagogical tool is provided, based on the author’s classroom experiences at three research universities and several conference-hosted tutorials.
An explicit, rule-based, category-learning task with abstract visual stimuli was administered to 50 healthy older adults and 48 younger adults. Accuracy and reaction time (RT) were examined for the effects of age, perceptual abilities, rule memory, rule complexity, stimulus novelty, and response competition. Older adults performed at equivalent levels to younger adults when applying a simple rule, but showed performance decrements when applying a more complex rule. The age effect interacted with both stimulus novelty and response competition, and was not eliminated after controlling for basic perceptual abilities and rule memory. The authors suggest that older adults show category learning deficits in conditions that require enhanced cognitive control. These results are discussed in reference to the growing body of literature regarding age-related change in executive abilities and frontal lobe function.
Computational models of cognition often exhibit complex dynamics that are difficult to discern without the use of visualization tools. Current tools often provide insight only to the modeling expert, however, and they provide limited functionality for communicating model dynamics to the nonexpert, as is needed during scientific presentations and in educational settings. We present NAV, the Node Activity Visualizer, an easy-to-use and portable software tool that interactively transforms the output of cognitive modeling simulators into presentation quality animations of model performance.
Human cognitive control is uniquely flexible and has been shown to depend on prefrontal cortex (PFC). But exactly how the biological mechanisms of the PFC support flexible cognitive control remains a profound mystery. Existing theoretical models have posited powerful task-specific PFC representations, but not how these develop. We show how this can occur when a set of PFC-specific neural mechanisms interact with breadth of experience to self organize abstract rule-like PFC representations that support flexible generalization in novel tasks. The same model is shown to apply to benchmark PFC tasks (Stroop and Wisconsin card sorting), accurately simulating the behavior of neurologically intact and frontally damaged people.
Many purported demonstrations of irrational behavior rely on the assumption that participants believe key task parameters that are merely asserted by experimenters. For example, previous researchers have found that participants who first reported confidence in items presented in a yes–no format did not change confidence to the degree prescribed by the normative model when those same items were later presented in a forced-choice format. A crucial assumption, however, was that participants fully believed the assertion that the forced-choice items were mutually exclusive and exhaustive. In this article, the authors derive and test a new normative model in which it is not assumed that participants fully believe the assertion. Two visual identification experiments show that the new normative model provides a compelling account of participants’ confidence reports.
We present a computational model of the intradimensional/ extradimensional (ID/ED) task (a variant of the Wisconsin card sorting task) that simulates the performance of intact and frontally lesioned monkeys on three different kinds of rule changes (Dias et al., 1997, J Neurosci 17:9285–9297). Although Dias et al. interpret the lesion data as supporting a model in which prefrontal cortex is organized into different processing functions, our model suggests an alternative account based on representational content. A key aspect of the model is that prefrontal cortex representations are organized according to different levels of abstraction, with orbital areas encoding more specific featural information and dorsolateral areas encoding more abstract dimensional information. This representational scheme of the model is integrated with two additional key elements: (i) activation-based working memory representations controlled by a dynamic gating mechanism that simulates the hypothesized phasic actions of dopaminergic neuromodulation in prefrontal cortex, which acts to stabilize or destabilize frontal representations based on success in the task; and (ii) a weight- based associative learning system simulating posterior cortex and other subcortical areas, where the stimulus–response mappings are encoded. Frontal cortex contributes to the task via top-down activation-based biasing of task-appropriate features and dimensions in this posterior cortex system — this top-down biasing is specifically important for overcoming prepotent associations after a sorting rule reverses. The ability of the model to capture the double-dissociation observed by Dias et al. with orbital versus dorsolateral lesions supports the validity of these principles, many of which have also been useful in accounting for other frontal phenomena.
Yes-no and forced-choice tasks are common in psychology, but the empirical relation between reported confidence in the 2 tasks has been unclear. The authors examined this relation with 2 experiments. The general experimental method had participants first report confidence in the truth of each of many general knowledge statements (a yes-no task) then report confidence in them again when the statements were put into pairs where it was known that one statement was true and one was false (a forced-choice task). At issue was how confidence in the statements changed between the yes-no task and the forced-choice task. Two models, including the normative one, were ruled out as descriptive models. A linear model and a multiplicative model remain viable contenders.