Constructing neural network models from brain data reveals representational transformations linked to adaptive behavior.

The human ability to adaptively implement a wide variety of tasks is thought to emerge from the dynamic transformation of cognitive information. We hypothesized that these transformations are implemented via conjunctive activations in "conjunction hubs"-brain regions that selectively integrate sensory, cognitive, and motor activations. We used recent advances in using functional connectivity to map the flow of activity between brain regions to construct a task-performing neural network model from fMRI data during a cognitive control task. We verified the importance of conjunction hubs in cognitive computations by simulating neural activity flow over this empirically-estimated functional connectivity model. These empirically-specified simulations produced above-chance task performance (motor responses) by integrating sensory and task rule activations in conjunction hubs. These findings reveal the role of conjunction hubs in supporting flexible cognitive computations, while demonstrating the feasibility of using empirically-estimated neural network models to gain insight into cognitive computations in the human brain.

Neural signatures of experience-based improvements in deterministic decision-making.

Feedback about our choices is a crucial part of how we gather information and learn from our environment. It provides key information about decision experiences that can be used to optimize future choices. However, our understanding of the processes through which feedback translates into improved decision-making is lacking. Using neuroimaging (fMRI) and cognitive models of decision-making and learning, we examined the influence of feedback on multiple aspects of decision processes across learning. Subjects learned correct choices to a set of 50 word pairs across eight repetitions of a concurrent discrimination task. Behavioral measures were then analyzed with both a drift-diffusion model and a reinforcement learning model. Parameter values from each were then used as fMRI regressors to identify regions whose activity fluctuates with specific cognitive processes described by the models. The patterns of intersecting neural effects across models support two main inferences about the influence of feedback on decision-making. First, frontal, anterior insular, fusiform, and caudate nucleus regions behave like performance monitors, reflecting errors in performance predictions that signal the need for changes in control over decision-making. Second, temporoparietal, supplementary motor, and putamen regions behave like mnemonic storage sites, reflecting differences in learned item values that inform optimal decision choices. As information about optimal choices is accrued, these neural systems dynamically adjust, likely shifting the burden of decision processing from controlled performance monitoring to bottom-up, stimulus-driven choice selection. Collectively, the results provide a detailed perspective on the fundamental ability to use past experiences to improve future decisions.

An fMRI study investigating effects of conceptually related sentences on the perception of degraded speech.

Prior research has shown that the perception of degraded speech is influenced by within sentence meaning and recruits one or more components of a frontal-temporal-parietal network. The goal of the current study is to examine whether the overall conceptual meaning of a sentence, made up of one set of words, influences the perception of a second acoustically degraded sentence, made up of a different set of words. Using functional magnetic resonance imaging (fMRI), we presented an acoustically clear sentence followed by an acoustically degraded sentence and manipulated the semantic relationship between them: Related in meaning (but consisting of different content words), Unrelated in meaning, or Same. Results showed that listeners' word recognition accuracy for the acoustically degraded sentences was significantly higher when the target sentence was preceded by a conceptually related compared to a conceptually unrelated sentence. Sensitivity to conceptual relationships was associated with enhanced activity in middle and inferior frontal, temporal, and parietal areas. In addition, the left middle frontal gyrus (LMFG), left inferior frontal gyrus (LIFG), and left middle temporal gyrus (LMTG) showed activity that correlated with individual performance on the Related condition. The superior temporal gyrus (STG) showed increased activation in the Same condition suggesting that it is sensitive to perceptual similarity rather than the integration of meaning between the sentence pairs. A fronto-temporo-parietal network appears to consolidate information sources across multiple levels of language (acoustic, lexical, syntactic, semantic) to build, and ultimately integrate conceptual information across sentences and facilitate the perception of a degraded speech signal. However, the nature of the sources of information that are available differentially recruit specific regions and modulate their activity within this network. Implications of these findings for the functional architecture of the network are considered.

Valuable Orientations Capture Attention.

Visual attention has long been known to be drawn to stimuli that are physically salient or congruent with task-specific goals. Several recent studies have shown that attention is also captured by stimuli that are neither salient nor task-relevant, but that are rendered in a color that has previously been associated with reward. We investigated whether another feature dimension-orientation-can be associated with reward via learning and thereby elicit value-driven attentional capture. In a training phase, participants received a monetary reward for identifying the color of Gabor patches exhibiting one of two target orientations. A subsequent test phase in which no reward was delivered required participants to search for Gabor patches exhibiting one of two spatial frequencies (orientation was now irrelevant to the task). Previously rewarded orientations robustly captured attention. We conclude that reward learning can imbue features other than color-in this case, specific orientations-with persistent value.

Evidence for Cerebellar Contributions to Adaptive Plasticity in Speech Perception.

Human speech perception rapidly adapts to maintain comprehension under adverse listening conditions. For example, with exposure listeners can adapt to heavily accented speech produced by a non-native speaker. Outside the domain of speech perception, adaptive changes in sensory and motor processing have been attributed to cerebellar functions. The present functional magnetic resonance imaging study investigates whether adaptation in speech perception also involves the cerebellum. Acoustic stimuli were distorted using a vocoding plus spectral-shift manipulation and presented in a word recognition task. Regions in the cerebellum that showed differences before versus after adaptation were identified, and the relationship between activity during adaptation and subsequent behavioral improvements was examined. These analyses implicated the right Crus I region of the cerebellum in adaptive changes in speech perception. A functional correlation analysis with the right Crus I as a seed region probed for cerebral cortical regions with covarying hemodynamic responses during the adaptation period. The results provided evidence of a functional network between the cerebellum and language-related regions in the temporal and parietal lobes of the cerebral cortex. Consistent with known cerebellar contributions to sensorimotor adaptation, cerebro-cerebellar interactions may support supervised learning mechanisms that rely on sensory prediction error signals in speech perception.

Value-driven attentional priority signals in human basal ganglia and visual cortex.

Goal-directed and stimulus-driven factors determine attentional priority through a well defined dorsal frontal-parietal and ventral temporal-parietal network of brain regions, respectively. Recent evidence demonstrates that reward-related stimuli also have high attentional priority, independent of their physical salience and goal-relevance. The neural mechanisms underlying such value-driven attentional control are unknown. Using human functional magnetic resonance imaging, we demonstrate that the tail of the caudate nucleus and extrastriate visual cortex respond preferentially to task-irrelevant but previously reward-associated objects, providing an attentional priority signal that is sensitive to reward history. The caudate tail has not been implicated in the control of goal-directed or stimulus-driven attention, but is well suited to mediate the value-driven control of attention. Our findings reveal the neural basis of value-based attentional priority.

Reward predictions bias attentional selection.

Attention selects stimuli for perceptual and cognitive processing according to an adaptive selection schedule. It has long been known that attention selects stimuli that are task relevant or perceptually salient. Recent evidence has shown that stimuli previously associated with reward persistently capture attention involuntarily, even when they are no longer associated with reward. Here we examine whether the capture of attention by previously reward-associated stimuli is modulated by the processing of current but unrelated rewards. Participants learned to associate two color stimuli with different amounts of reward during a training phase. In a subsequent test phase, these previously rewarded color stimuli were occasionally presented as to-be-ignored distractors while participants performed visual search for each of two differentially rewarded shape-defined targets. The results reveal that attentional capture by formerly rewarded distractors was the largest when both recently received and currently expected reward were the highest in the test phase, even though such rewards were unrelated to the color distractors. Our findings support a model in which value-driven attentional biases acquired through reward learning are maintained via the cognitive mechanisms involved in predicting future rewards.

A Neural Mechanism for Reward Discounting: Insights from Modeling Hippocampal-Striatal Interactions.

Decision-making often requires taking into consideration immediate gains as well as delayed rewards. Studies of behavior have established that anticipated rewards are discounted according to a decreasing hyperbolic function. Although mathematical explanations for reward delay discounting have been offered, little has been proposed in terms of neural network mechanisms underlying discounting. There has been much recent interest in the potential role of the hippocampus. Here we demonstrate that a previously-established neural network model of hippocampal region CA3 contains a mechanism that could explain discounting in downstream reward-prediction systems (e.g., basal ganglia). As part of its normal function, the model forms codes for stimuli that are similar to future, predicted stimuli. This similarity provides a means for reward predictions associated with future stimuli to influence current decision-making. Simulations show that this "predictive similarity" decreases as the stimuli are separated in time, at a rate that is consistent with hyperbolic discounting.

Rapid instructed task learning: a new window into the human brain's unique capacity for flexible cognitive control.

The human ability to flexibly adapt to novel circumstances is extraordinary. Perhaps the most illustrative, yet underappreciated, form of this cognitive flexibility is rapid instructed task learning (RITL)--the ability to rapidly reconfigure our minds to perform new tasks from instructions. This ability is important for everyday life (e.g., learning to use new technologies) and is used to instruct participants in nearly every study of human cognition. We review the development of RITL as a circumscribed domain of cognitive neuroscience investigation, culminating in recent demonstrations that RITL is implemented via brain circuits centered on lateral prefrontal cortex. We then build on this and the recent discovery of compositional representations within lateral prefrontal cortex to develop an integrative theory of cognitive flexibility and cognitive control that identifies mechanisms that may enable RITL within the human brain. The insights gained from this new theoretical account have important implications for further developments and applications of RITL research.

Reward and attentional control in visual search.

It has long been known that the control of attention in visual search depends both on voluntary, top-down deployment according to context-specific goals, and on involuntary, stimulus-driven capture based on the physical conspicuity of perceptual objects. Recent evidence suggests that pairing target stimuli with reward can modulate the voluntary deployment of attention, but there is little evidence that reward modulates the involuntary deployment of attention to task-irrelevant distractors. We report several experiments that investigate the role of reward learning on attentional control. Each experiment involved a training phase and a test phase. In the training phase, different colors were associated with different amounts of monetary reward. In the test phase, color was not task-relevant and participants searched for a shape singleton; in most experiments no reward was delivered in the test phase. We first show that attentional capture by physically salient distractors is magnified by a previous association with reward. In subsequent experiments we demonstrate that physically inconspicuous stimuli previously associated with reward capture attention persistently during extinction--even several days after training. Furthermore, vulnerability to attentional capture by high-value stimuli is negatively correlated across individuals with working memory capacity and positively correlated with trait impulsivity. An analysis of intertrial effects reveals that value-driven attentional capture is spatially specific. Finally, when reward is delivered at test contingent on the task-relevant shape feature, recent reward history modulates value-driven attentional capture by the irrelevant color feature. The influence of learned value on attention may provide a useful model of clinical syndromes characterized by similar failures of cognitive control, including addiction, attention-deficit/hyperactivity disorder, and obesity.

Generalization of value-based attentional priority.

Attention is the mechanism by which important or salient stimuli are selected for perceptual and cognitive processing. Which stimuli are attended has important implications for effective goal-directed behaviour, survival, and well-being. A growing body of evidence suggests that reward-predicting stimuli capture attention involuntarily. In previous studies, value-based attentional priority has been observed only when the formerly reward-related stimuli themselves were presented as targets or distractors. Here we show that stimulus-reward associations learned in one task generalize to different stimuli that share a defining feature (colour) in another task. Our results reveal a broad and flexible role for reward learning in modulating attentional priority.

Learned value magnifies salience-based attentional capture.

Visual attention is captured by physically salient stimuli (termed salience-based attentional capture), and by otherwise task-irrelevant stimuli that contain goal-related features (termed contingent attentional capture). Recently, we reported that physically nonsalient stimuli associated with value through reward learning also capture attention involuntarily (Anderson, Laurent, & Yantis, PNAS, 2011). Although it is known that physical salience and goal-relatedness both influence attentional priority, it is unknown whether or how attentional capture by a salient stimulus is modulated by its associated value. Here we show that a physically salient, task-irrelevant distractor previously associated with a large reward slows visual search more than an equally salient distractor previously associated with a smaller reward. This magnification of salience-based attentional capture by learned value extinguishes over several hundred trials. These findings reveal a broad influence of learned value on involuntary attentional capture.

Value-driven attentional capture.

Attention selects which aspects of sensory input are brought to awareness. To promote survival and well-being, attention prioritizes stimuli both voluntarily, according to context-specific goals (e.g., searching for car keys), and involuntarily, through attentional capture driven by physical salience (e.g., looking toward a sudden noise). Valuable stimuli strongly modulate voluntary attention allocation, but there is little evidence that high-value but contextually irrelevant stimuli capture attention as a consequence of reward learning. Here we show that visual search for a salient target is slowed by the presence of an inconspicuous, task-irrelevant item that was previously associated with monetary reward during a brief training session. Thus, arbitrary and otherwise neutral stimuli imbued with value via associative learning capture attention powerfully and persistently during extinction, independently of goals and salience. Vulnerability to such value-driven attentional capture covaries across individuals with working memory capacity and trait impulsivity. This unique form of attentional capture may provide a useful model for investigating failures of cognitive control in clinical syndromes in which value assigned to stimuli conflicts with behavioral goals (e.g., addiction, obesity).

The emergence of saliency and novelty responses from Reinforcement Learning principles.

Recent attempts to map reward-based learning models, like Reinforcement Learning [Sutton, R. S., & Barto, A. G. (1998). Reinforcement Learning: An introduction. Cambridge, MA: MIT Press], to the brain are based on the observation that phasic increases and decreases in the spiking of dopamine-releasing neurons signal differences between predicted and received reward [Gillies, A., & Arbuthnott, G. (2000). Computational models of the basal ganglia. Movement Disorders, 15(5), 762-770; Schultz, W. (1998). Predictive reward signal of dopamine neurons. Journal of Neurophysiology, 80(1), 1-27]. However, this reward-prediction error is only one of several signals communicated by that phasic activity; another involves an increase in dopaminergic spiking, reflecting the appearance of salient but unpredicted non-reward stimuli [Doya, K. (2002). Metalearning and neuromodulation. Neural Networks, 15(4-6), 495-506; Horvitz, J. C. (2000). Mesolimbocortical and nigrostriatal dopamine responses to salient non-reward events. Neuroscience, 96(4), 651-656; Redgrave, P., & Gurney, K. (2006). The short-latency dopamine signal: A role in discovering novel actions? Nature Reviews Neuroscience, 7(12), 967-975], especially when an organism subsequently orients towards the stimulus [Schultz, W. (1998). Predictive reward signal of dopamine neurons. Journal of Neurophysiology, 80(1), 1-27]. To explain these findings, Kakade and Dayan [Kakade, S., & Dayan, P. (2002). Dopamine: Generalization and bonuses. Neural Networks, 15(4-6), 549-559.] and others have posited that novel, unexpected stimuli are intrinsically rewarding. The simulation reported in this article demonstrates that this assumption is not necessary because the effect it is intended to capture emerges from the reward-prediction learning mechanisms of Reinforcement Learning. Thus, Reinforcement Learning principles can be used to understand not just reward-related activity of the dopaminergic neurons of the basal ganglia, but also some of their apparently non-reward-related activity.

Serial or parallel? Using depth-of-processing to examine attention allocation during reading.

This paper presents an experiment investigating attention allocation in four tasks requiring varied degrees of lexical processing of 1-4 simultaneously displayed words. Response times and eye movements were only modestly affected by the number of words in an asterisk-detection task but increased markedly with the number of words in letter-detection, rhyme-judgment, and semantic-judgment tasks, suggesting that attention may not be serial for tasks that do not require significant lexical processing (e.g., detecting visual features), but is approximately serial for tasks that do (e.g., retrieving word meanings). The implications of these results for models of readers' eye movements are discussed.

Using reinforcement learning to understand the emergence of "intelligent" eye-movement behavior during reading.

The eye movements of skilled readers are typically very regular (K. Rayner, 1998). This regularity may arise as a result of the perceptual, cognitive, and motor limitations of the reader (e.g., limited visual acuity) and the inherent constraints of the task (e.g., identifying the words in their correct order). To examine this hypothesis, reinforcement learning was used to allow an artificial "agent" to learn to move its eyes to read as efficiently as possible. The resulting patterns of simulated eye movements resembled those of skilled readers and suggest that important aspects of eye-movement behavior might emerge as a consequence of satisfying the constraints that are imposed on readers. These results also suggest novel interpretations of some contentious empirical results, such as the fixation duration costs associated with word skipping (R. Kliegl & R. Engbert, 2005), and theoretical assumptions, for example the familiarity check in the E-Z Reader model of eye-movement control (E. D. Reichle, A. Pollatsek, D. L. Fisher, & K. Rayner, 1998).