Learning the exception to the rule: model-based FMRI reveals specialized representations for surprising category members

Category knowledge can be explicit, yet not conform to a perfect rule. For example, a child may acquire the rule "If it has wings, then it is a bird," but then must account for exceptions to this rule, such as bats. The current study explored the neurobiological basis of rule-plus-exception learning by using quantitative predictions from a category learning model, SUSTAIN, to analyze behavioral and functional magnetic resonance imaging (fMRI) data. SUSTAIN predicts that exceptions require formation of specialized representations to distinguish exceptions from rule-following items in memory. By incorporating quantitative trial-by-trial predictions from SUSTAIN directly into fMRI analyses, we observed medial temporal lobe (MTL) activation consistent with 2 predicted psychological processes that enable exception learning: item recognition and error correction. SUSTAIN explains how these processes vary in the MTL across learning trials as category knowledge is acquired. Importantly, MTL engagement during exception learning was not captured by an alternate exemplar-based model of category learning or by standard contrasts comparing exception and rule-following items. The current findings thus provide a well-specified theory for the role of the MTL in category learning, where the MTL plays an important role in forming specialized category representations appropriate for the learning context.     

Medial temporal lobe involvement in an implicit memory task: evidence of collaborating implicit and explicit memory systems from FMRI and Alzheimer's disease

We used a prototype extraction task to assess implicit learning of a meaningful novel visual category. Cortical activation was monitored in young adults with functional magnetic resonance imaging. We observed occipital deactivation at test consistent with perceptually based implicit learning, and lateral temporal cortex deactivation reflecting implicit acquisition of the category's semantic nature. Medial temporal lobe (MTL) activation during exposure and test suggested involvement of explicit memory as well. Behavioral performance of Alzheimer's disease (AD) patients and healthy seniors was also assessed, and AD performance was correlated with gray matter volume using voxel-based morphometry. AD patients showed learning, consistent with preserved implicit memory, and confirming that AD patients' implicit memory is not limited to abstract patterns. However, patients were somewhat impaired relative to healthy seniors. Occipital and lateral temporal cortical volume correlated with successful AD patient performance, and thus overlapped with young adults' areas of deactivation. Patients' severe MTL atrophy precluded involvement of this region. AD patients thus appear to engage a cortically based implicit memory mechanism, whereas their relative deficit on this task may reflect their MTL disease. These findings suggest that implicit and explicit memory systems collaborate in neurologically intact individuals performing an ostensibly implicit memory task.     

Dissociation between explicit memory and configural memory in the human medial temporal lobe

Using functional magnetic resonance imaging, the current study explored the differential mnemonic contributions of the hippocampus and surrounding medial temporal lobe (MTL) cortices to explicit recognition memory and configural learning. Using a task that required processing of repeated and novel visuospatial contexts across multiple trials, we examined MTL activation in relation to 3 forms of learning in a single paradigm: 1) context-independent procedural learning, 2) context-dependent configural learning, and 3) explicit recognition memory. Activations in hippocampus and parahippocampal cortex were associated with explicit memory, differentiating between subsequently remembered and forgotten repeated contexts, but were unrelated to context-dependent configural learning. Activations in regions of perirhinal and entorhinal cortex were associated with configural learning of repeated contexts independent from explicit memory for those contexts. Procedural learning was unrelated to activation in any MTL region. The time course of activation across learning further differed in MTL subregions with MTL cortex demonstrating repetition-related decreases and hippocampus repetition-related increases. These repetition effects were differentially sensitive to recognition with only activation in hippocampus and parahippocampal cortex tracking recognized items. These imaging findings converge with studies of amnesia and indicate dissociable roles for hippocampus in learning that supports explicit recognition and for anterior MTL cortex in configural learning.     

Learning-related neuronal responses in prefrontal cortex studied with functional neuroimaging

We assessed time-dependent neuronal activity accompanying learning using functional magnetic resonance imaging (fMRI). An artificial grammar learning paradigm enabled us to dissociate activations associated with individual item learning from those involved in learning the underlying grammar system. We show that a localized region of right prefrontal cortex (PFC) is preferentially sensitive to individual item learning during the early stages of the experiment, while the left PFC region is sensitive to grammar learning which occurred across the entire course of the experiment. In addition to dissociating these two types of learning, we were able to characterize the effect of rule acquisition on neuronal responses associated with explicit learning of individual items. This effect was expressed as modulation of the time-dependent right PFC activations such that the early increase in activation associated with item learning was attenuated as the experiment progressed. In a further analysis we used structural equation modelling to explore time-dependent changes in inter-regional connectivity as a function of both item and grammar rule learning. Although there were no significant effects of item learning on the measured path strengths, rule learning was associated with a decrease in right fronto-parietal connectivity and an increase in connectivity between left and right PFC. Further fronto-parietal path strengths were observed to change, with an increase in left fronto-parietal and a decrease in right fronto-parietal connectivity path strength from right PFC to left parietal cortex. We interpret our findings in terms of a left frontal system mediating the semantic analysis of study items and directly influencing a right fronto-parietal system associated with episodic memory retrieval.     

Two distinct neural mechanisms for category-selective responses

The cognitive and neural mechanisms mediating category-selective responses in the human brain remain controversial. Using functional magnetic resonance imaging and effective connectivity analyses (Dynamic Causal Modelling), we investigated animal- and tool-selective responses by manipulating stimulus modality (pictures versus words) and task (implicit versus explicit semantic). We dissociated two distinct mechanisms that engender category selectivity: in the ventral occipito-temporal cortex, tool-selective responses were observed irrespective of task, greater for pictures and mediated by bottom-up effects. In a left temporo-parietal action system, tool-selective responses were observed irrespective of modality, greater for explicit semantic tasks and mediated by top-down modulation from the left prefrontal cortex. These distinct activation and connectivity patterns suggest that the two systems support different cognitive operations, with the ventral occipito-temporal regions engaged in structural processing and the dorsal visuo-motor system in strategic semantic processing. Consistent with current semantic theories, explicit semantic processing of tools might thus rely on reactivating their associated action representations via top-down modulation. In terms of neuronal mechanisms, the category selectivity may be mediated by distinct top-down (task-dependent) and bottom-up (stimulus-dependent) mechanisms.     

An event-related fMRI study of explicit syntactic processing of normal/anomalous sentences in contrast to implicit syntactic processing

Using event-related functional magnetic resonance imaging (fMRI), we examined activation of cortical language areas for explicit syntactic processing. In a syntactic decision (Syn) task, the participants judged whether the presented sentence was syntactically correct, where syntactic knowledge about the distinction between transitive and intransitive verbs was required. In a semantic decision (Sem) task, lexico-semantic knowledge about selectional restrictions was indispensable. In a phonological decision (Pho) task, phonological knowledge about accent patterns was required. The Sem and Pho tasks involved implicit syntactic processing, as well as explicit semantic and phonological processing, respectively. We also tested a voice-pitch comparison (Voi) task in which no explicit linguistic knowledge was required. In the direct comparison of Syn - (Sem + Pho + Voi), we found localized activation in the left inferior frontal gyrus (F3op/F3t), indicating that activation of the left F3op/F3t is more prominently enhanced in explicit syntactic processing than in implicit syntactic processing. Moreover, we determined that its activation is selective to syntactic judgments regarding both normal and anomalous sentences. These results suggest that explicit information processing in the syntactic domain critically involves the left F3op/F3t, which is functionally separable from other regions.     

On the benefits of not trying: brain activity and connectivity reflecting the interactions of explicit and implicit sequence learning

Under certain circumstances, implicit, automatic learning may be attenuated by explicit memory processes. We explored the brain basis of this phenomenon in a functional magnetic resonance imaging (fMRI) study of motor sequence learning. Using a factorial design that crossed subjective intention to learn (explicit versus implicit) with sequence difficulty (a standard versus a more complex alternating sequence), we show that explicit attempts to learn the difficult sequence produce a failure of implicit learning and, in a follow-up behavioural experiment, that this failure represents a suppression of learning itself rather than of the expression of learning. This suppression is associated with sustained right frontal activation and attenuation of learning-related changes in the medial temporal lobe and the thalamus. Furthermore, this condition is characterized by a reversal of the fronto-thalamic connectivity observed with unimpaired implicit learning. The findings demonstrate a neural basis for a well-known behavioural effect: the deleterious impact of an explicit search upon implicit learning.     

Distinctive neural processes during learning in autism

This functional magnetic resonance imaging study compared the neural activation patterns of 18 high-functioning individuals with autism and 18 IQ-matched neurotypical control participants as they learned to perform a social judgment task. Participants learned to identify liars among pairs of computer-animated avatars uttering the same sentence but with different facial and vocal expressions, namely those that have previously been associated with lying versus truth-telling. Despite showing a behavioral learning effect similar to the control group, the autism group did not show the same pattern of decreased activation in cortical association areas as they learned the task. Furthermore, the autism group showed a significantly smaller increase in interregion synchronization of activation (functional connectivity) with learning than did the control group. Finally, the autism group had decreased structural connectivity as measured by corpus callosum size, and this measure was reliably related to functional connectivity measures. The findings suggest that cortical underconnectivity in autism may constrain the ability of the brain to rapidly adapt during learning.     

Comparing the effects of stimulation and propofol infusion rate on implicit and explicit memory formation

Doubt remains about the conditions under which learning persistsdespite anaesthesia. This study investigated the relative importanceof dose of anaesthetic and stimulation for learning during propofolinfusion before surgery. Thirty-six patients were randomly assignedto three groups. Group 1 received two word lists (category examplesand nonsense words) during infusion of propofol to a targetconcentration of 2 µg ml^–1. Groups 2 and3 received the word lists during infusion of propofol 5 µg ml^–1.Group 2 received nonsense words before tracheal intubation andcategory examples during intubation; Group 3 heard categoryexamples before and nonsense words during intubation. Bispectralindex was recorded as a measure of depth of sedation/anaesthesia.We assessed explicit memory on recovery using a structured interviewand a recognition test. We assessed implicit memory using acategory generation test and a preference rating task. To establishbaseline, a control group of 12 patients completed the categorygeneration test without receiving the category examples duringanaesthesia. Overall, there was no evidence for learning duringpropofol infusion, though the category generation task showeda trend towards more implicit memory for words presented duringintubation than during anaesthesia. We conclude that learningdoes not occur during anaesthesia without surgery. Br J Anaesth 2001; 86: 189–95     

Neurocomputational Changes in Inhibitory Control Associated With Prolonged Exposure Therapy

Posttraumatic stress disorder (PTSD) is associated with inhibitory control dysfunction that extends beyond difficulties inhibiting trauma-related intrusions. Inhibitory learning has been proposed as a potential mechanism of change underlying the effectiveness of extinction-based therapies such as prolonged exposure (PE), a first-line treatment for PTSD. To identify neurocognitive markers of change in inhibitory learning associated with PE, we applied a Bayesian learning model to the analysis of neuroimaging data collected during an inhibitory control task, both before and after PE treatment. Veterans (N = 20) with combat-related PTSD completed a stop-signal task (SST) while undergoing fMRI at time points immediately before and after PE treatment. Participants exhibited a small, significant improvement in performance on the SST, as demonstrated by longer reaction times and improved inhibition accuracy. Amplitude of neural activation associated with a signed prediction error (SPE; i.e., the discrepancy between actual outcome and model-based expectation of needing to stop) in the right caudate decreased from baseline to posttreatment assessment. Change in model-based activation was modulated by performance accuracy, with a decrease in positive SPE activation observed on successful trials, d = 0.79, and a reduction in negative SPE activation on error trials, d = 0.74. The decrease in SPE-related activation on successful stop trials was correlated with PTSD symptom reduction. These results are consistent with the notion that PE may help broadly strengthen inhibitory learning and the development of more accurate model-based predictions, which may thus facilitate change in cognitions in response to trauma-related cues and help reduce PTSD symptoms.     

Regional brain activation associated with different performance patterns during learning of a complex motor skill

In understanding the brain's response to extensive practice and development of high-level, expert skill, a key question is whether the same brain structures remain involved throughout the different stages of learning and a form of adaptation occurs, or a new functional circuit is formed with some structures dropping off and others joining. After training subjects on a set of complex motor tasks (tying knots), we utilized fMRI to observe that in subjects who learned the task well new regional activity emerged in posterior medial structures, i.e. the posterior cingulate gyrus. Activation associated with weak learning of the knots involved areas that mediate visual spatial computations. Brain activity associated with no substantive learning indicated involvement of areas dedicated to the declarative aspects learning such as the anterior cingulate and prefrontal cortex. The new activation for the pattern of strong learning has alternate interpretations involving either retrieval during episodic memory or a shift toward non-executive cognitive control of the task. While these interpretations are not resolved, the study makes clear that single time-point images of motor skill can be misleading because the brain structures that implement action can change following practice.     

Regional brain activation during concurrent implicit and explicit sequence learning

We used event-related fMRI to identify the brain regions engaged during explicit and implicit sequence learning (ESL and ISL, respectively). Twenty-four subjects performed a concurrent ESL and ISL task. Behavior showed learning in both conditions. Prefrontal (PFC), striatal, anterior cingulate cortex (ACC) and visual regions (V1, V2 and V3) were engaged during both ESL and ISL. With ESL there was increased activity in the visual regions on the predictable (i.e. learned pattern) trials. With ISL, however, there was a relative decrease in activity in visual regions. The opposite patterns in the visual regions highlight the different effects of ESL and ISL. The learning process was distinguished from the result of learning, by fitting subjects' functional magnetic resonance imaging data to their learning curve. This analysis revealed more extensive PFC activity during ESL and caudal ACC activity specific for the result of learning analysis, when the expected response was violated. Our results suggest a relative dissociation of the brain regions engaged during ESL and ISL, whereby ESL and ISL can be viewed as partially distinct but overlapping parallel processes.     

The functional anatomy of sleep-dependent visual skill learning

Learning of procedural skills develops gradually, with performance improving significantly with practice. But improvement on some tasks, including a visual texture discrimination task, continues in the absence of further practice, expressly during periods of sleep and not across equivalent waking episodes. Here we report that the brain activation revealed significantly different patterns of performance-related functional activity following a night of sleep relative to 1 h post-training without intervening sleep. When task activation patterns after a night of sleep were compared with activation patterns without intervening sleep (1 h post-training), significant regions of increased signal intensity were observed in the primary visual cortex, the occipital temporal junction, the medial temporal lobe and the inferior parietal lobe. In contrast, a region of decreased signal intensity was found in the right temporal pole. Corroborating these condition differences, correlations between behavioural performance and brain activation revealed significantly different patterns of performance-related functional activity following a night of sleep relative to those without intervening sleep. Together, these data provide evidence of overnight bi-directional changes in functional anatomy, differences that may form the neural basis of sleep-dependent learning expressed on this task.     

The impact of pathology review on treatment recommendations for patients with adenocarcinoma of the prostate

This study was designed to estimate the frequency with which changes in Gleason score because of a genitourinary pathologist's review changed prostate cancer treatment recommendations. The study cohort consisted of 602 patients who presented to a genitourinary oncologist for a second opinion after being diagnosed with prostate cancer based on a needle biopsy at a nonacademic institution from 1989 through 2001. Each of the prostate biopsy specimens was sent for review by a genitourinary pathologist. Applying the rule that low-risk patients would receive monotherapy, and intermediate or high-risk patients would receive combined modality therapy, the frequency with which treatment recommendations were changed by pathology review was calculated. Pathology review by a genitourinary pathologist changed the Gleason score by at least 1 point in 44% of cases. Upgrades were more common than downgrades and accounted for 81% [95% confidence interval: 76-86%] of the changes. Patients' risk category was increased in 10.8% of cases and was decreased in 3.4%. Risk category was changed from low risk to intermediate or high risk in 8.2%, but was changed from intermediate or high risk to low risk in only 0.9%. Genitourinary pathology review would have changed management in approximately 10% of men, mainly in the direction of combined therapy over monotherapy.     

The role of the dorsolateral prefrontal cortex during sequence learning is specific for spatial information.

Many studies have implicated the dorsolateral prefrontal cortex in the acquisition of skill, including procedural sequence learning. However, the specific role it performs in sequence learning has remained uncertain. This type of skill has been intensively studied using the serial reaction time task. We used three versions of this task: a standard task where the position of the stimulus cued the response; a non-standard task where the color of the stimulus was related to the correct response; and a combined task where both the color and position simultaneously cued the response. We refer to each of these tasks based upon the cues available for guiding learning as position, color and combined tasks. The combined task usually shows an enhancement of skill acquisition, a result of being driven by two simultaneous and congruent cues. Prior to the performance of each of these tasks the function of the dorsolateral prefrontal cortex was disrupted using repetitive transcranial magnetic stimulation. This completely prevented learning within the position task, while sequence learning occurred to a similar extent in both the color and combined tasks. So, following prefrontal stimulation the expected learning enhancement in the combined task was lost, consistent with only a color cue being available to guide sequence learning in the combined task. Neither of these effects was observed following stimulation at the parietal cortex. Hence the critical role played by the dorsolateral prefrontal cortex in sequence learning is related exclusively to spatial cues. We suggest that the dorsolateral prefrontal cortex operates over the short term to retain and manipulate spatial information to allow cortical and subcortical structures to learn a predictable sequence of actions. Such functions may emerge from the broader role the dorsolateral prefrontal cortex has in spatial working memory. These results argue against the dorsolateral prefrontal cortex constituting part of the neuronal substrate responsible for general aspects of implicit or explicit sequence learning.     

EFFECTS OF LOW CONCENTRATIONS OF ENFLURANE ON PROBABILITY LEARNING

The effects of controlled subanaesthetic concentrations of enfluraneon learning behaviour and on ability to change previously developeddecision strategies were studied in 10 male volunteers, usinga probability learning task. Subjects were instructed to predicton each of 200 consecutive trials, whether a left or a rightlight would appear. The appearance of lights was pre-determinedby a set relative frequency unknown to the subject. The relativefrequency was automatically changed at the end of the first100 trials, from 8/10 lights in the left side to 4/10. It wasfound that enflurane at subanaesthetic concentration of 0.25%(end-tidal) slowed the rate of learning, and increased the numberof trials required for readjusting the prediction strategy tothe changed situation. ^*Present address (for correspondence): Aranne Laboratory ofHuman Neuropsychology, Department of Neurology, Hadassah UniversityHospital, Jerusalem, Israel.     

Neural evidence for dissociable components of task-switching

The ability to retrieve and flexibly switch between task rules is seen as an important component of cognitive control. It is often assumed that lateral prefrontal cortex (latPFC) is important for switching between rules. However, activation associated with rule-switching is less reliably observed in latPFC than in medial PFC (specifically, pre-supplementary motor area). In this study, we tested the hypothesis that medial PFC is important for reconfiguration of task sets, whereas latPFC is important for retrieving, maintaining and implementing relevant rules (i.e. rule representation). Twenty young adults participated in a functional magnetic resonance imaging study in which they determined the correct response to a target stimulus on the basis of an instructional cue. For bivalent targets, the appropriate response depended on the currently relevant rule. In contrast, univalent targets were always associated with the same response. Brain regions of interest were characterized according to their responsiveness to bivalent and univalent targets, on both rule-switch and rule-repetition trials. The data support the hypothesis that rule representation and task-set reconfiguration are separable cognitive processes, associated with dissociable neural activation in latPFC and medial PFC, respectively. Activation profiles of posterior parietal cortex, basal ganglia and rostrolateral PFC are also examined and discussed.     

A link between the hippocampal and the striatal memory systems of the brain

Two major memory systems have been recognized over the years (Squire 1987): the declarative memory system, which is under the control of the hippocampus and related temporal lobe structures, and the procedural or habit memory system, which is under the control of the striatum and its connections. Most if not all learning tasks studied in animals, however, involve either the performance or the suppression of movement; this, if learned well, may be viewed as having become a habit. It is agreed that memory rules change from their first association to those that take place when the task is mastered. Does this change of rules involve a switch from one memory system to another? Here we will comment on: 1) reversal learning in the Morris water maze (MWM), in which the declarative or spatial component of a task is changed but the procedural component (to swim to safety) persists and needs to be re-linked with a different set of spatial cues; and 2) a series of observations on an inhibitory avoidance task that indicate that the brain systems involved change with further learning.     

Functional neuroanatomy of anticipatory behavior: dissociation between sensory-driven and memory-driven systems

The ability to anticipate predictable stimuli allows faster responses. The predictive saccade (PRED) task has been shown to quickly induce such anticipatory behavior in humans. In a PRED task subjects track a visual target jumping back and forth between fixed positions at a fixed time interval. During this task, saccade latencies drop from approximately 200 ms to <80 ms as subjects anticipate target appearance. This change in saccade latency indicates that subjects' behavior shifts from being sensory driven to being memory driven. We conducted functional magnetic resonance imaging studies with 10 healthy adults performing the PRED task using a standard block design. We compared the PRED task with a visually guided saccade (VGS) task using unpredictable targets matched for number, direction and amplitude of required saccades. Our results show greater activation during the PRED task in the prefrontal, pre-supplementary motor and anterior cingulate cortices, hippocampus, mediodorsal thalamus, striatum and cerebellum. The VGS task elicited greater activation in the cortical eye fields and occipital cortex. These results demonstrate the important dissociation between sensory and predictive neural control of similar saccadic eye movements. Anticipatory behavior induced by the PRED task required less sensory-related processing activity and was subserved by a distributed cortico-subcortical memory system including prefronto-striatal circuitry.     

A cellular correlate of learning-induced metaplasticity in the hippocampus

Metaplasticity, the plasticity of synaptic plasticity, is thought to have a pivotal role in activity-dependent modulation of synaptic connectivity, which underlies learning and memory. Metaplasticity is usually attributed to modifications in glutamate receptor-mediated synaptic transmission. However, experimental evidence and theoretical considerations suggest that learning reduces the predisposition for further synaptic strengthening, while behavioral studies show that learning capability is enhanced by prior learning. Here we show that enhanced neuronal excitability in CA1 pyramidal neurons, but not enhanced synaptic transmission, occurs prior to rule learning of an olfactory discrimination task. This transient enhancement lasts for 1 day after rule learning, is apparent throughout the cell population and results from reduction in the medium and slow after-hyperpolarizations that control spike frequency adaptation. Such olfactory learning-induced increased excitability in hippocampal neurons enhances the rats' learning capability in another hippocampus-dependent task, the Morris water maze. Once olfactory discrimination rule learning is acquired, its maintenance is not dependent on the reduced post-burst AHP in hippocampal neurons. However, the enhanced spatial learning capability of olfactory-trained rats in the water maze is diminished once the post burst AHP in CA1 pyramidal cells resumes its initial value. We suggest that enhanced excitability of CA1 neurons may serve as a mechanism for generalized enhancement of hippocampus-dependent learning capability. In the presence of such enhanced neuronal excitability, the hippocampal network enters into a 'learning mode' in which a variety of hippocampus-dependent skills are acquired rapidly and efficiently.