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.generalization ͉ abstraction ͉ adaptive gating A fundamental human cognitive faculty is the capacity for cognitive control: the ability to behave in accord with rules, goals, or intentions, even when this runs counter to reflexive or otherwise highly compelling competing responses (e.g., the ability to keep typing rather than scratch a mosquito bite). A hallmark of cognitive control in humans is its remarkable flexibility: we can perform novel tasks with very little additional experience (e.g., playing a card game for the first time by observing the play or hearing the rules described). This ability appears to depend on the prefrontal cortex (PFC) (1-5) and in particular on abstract rule-like representations localized to this brain area (6-8). However, this capacity emerges only slowly over a protracted period through late adolescence, closely tracking the development of the PFC (9-11). At the psychological level, flexible cognitive control has been modeled abstractly in terms of symbol processing computations that support arbitrary variable binding (12). However, it remains unclear whether or how such models correspond to the increasingly rich body of knowledge about the neural mechanisms underlying cognitive control and in particular the functioning of the PFC. At the biological level, a number of neural models have proposed that cognitive control relies on the active maintenance of abstract rule-like representations in PFC that guide processing in posterior cortex (13-17). However, none of these existing frameworks have explained how such representations might develop, and why this development should take so long; indeed, most models rely on hand-coded representations designed explicitly for solving a specific set of tasks. Thus, a major challenge to theories of the neural bases of cognitive control remains unanswered: how it can be explained in terms of self-organizing mechanisms that develop on their own, over time, without recourse to unexplained sources of influence or intelligence (i.e., a ''homunculus'') (18).Here, we present a computational model that provides an explanation for the development of cognitive flexibility. This model shows how neurobiological mechanisms specific to the PFC result in the self organization of abstract rule-like PFC representations that support flexible cognit...

Connectionist and dynamical systems approaches explain human thought, language and behavior in terms of the emergent consequences of a large number of simple non-cognitive processes. We view the entities that serve as the basis for structured probabilistic approaches as sometimes useful but often misleading abstractions that have no real basis in the actual processes that give rise to linguistic and cognitive abilities or the development of these abilities. While structured probabilistic approaches can be useful in determining what would be optimal under certain assumptions, we suggest that approaches such as the connectionist and dynamical systems approaches, which focus on explaining the mechanisms giving rise to cognition, will be essential in achieving a full understanding of cognition and development.

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.

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.ne of the most impressive aspects of human cognition is also one of its most enduring mysteries: how it can respond in appropriate ways to novel circumstances. In our everyday lives, we are constantly confronted with the need to make sense of and respond appropriately to new situations. Almost always, the individual constituents of these situations (e.g., the people, places, and/or actions involved) are things with which we have had prior experience, and it is the particular combination that is new. A person may appear in a new context or carry out an action we have never before witnessed them perform, or a word may be used in a novel way within a sentence. Nevertheless, we are able to makes sense of and respond appropriately to such circumstances, drawn from a nearly infinite array of possible combinations, despite having had experience with only a limited number of them. It has frequently been argued that this flexibility, or systematicity, relies on symbol processing, that is, the ability to represent information in the form of abstract variables that can be bound to arbitrary values, as is possible in a symbol system. For example, in trying to understand a sentence, if the constituent parts can be represented as variables, then any possible word can be assigned, or "bound," to each (e.g., in the sentence "I want to desk you," "desk" can be understood as the verb). Such variable binding provides tremendous flexibility and is fundamental to the power of computer systems. However, whether and how this ability is implemented in the brain remains one of the great mysteries of neuroscience. Historically, this ability has been used as a key argument by those advocating for symbolic cognitive models, over "associationist" neural network models (1, 2). In response, some have argued that human symbol processing ability is limited at best and that many...