Search

A defi ning trait of cognition is the capacity to combine information into compound concepts. This ability relies, among others, on the logical connectives ‘and’, ‘or’ and ‘if-then’. Simple sentences, such as “there is a fork on the table” (A) or “there is a knife” (B), can be combined in different ways using different connectives. No evidence is available to date on how and where the brain represents different concept combinations produced by different connectives, and how these are evaluated against new facts. Here, participants learned associations between graphic cues and conjunctive (A and B), disjunctive (A or B) or conditional (If A then B) sentences. During fMRI scanning, a cue was presented, followed by a delay, during which participants had to represent the sentence associated to the cue; fi nally, a visual scene had to be evaluated for compatibility with the sentence. Two participant groups were recruited so that conditionals (If A then B) were interpreted in either of two alternative ways (thus, same form, different semantics). Multivariate decoding applied to the delay period revealed that the active sentence was encoded in left inferior frontal gyrus (BA44). During the delay, no difference was found between participant groups. During the target phase, we found higher activations in rostral regions of left inferior frontal cortex (BA47), for disjunctions and conditionals relative to conjunctions. Activation of the inferior parietal lobe only was modulated by the interpretation of conditionals.

The advent of functional magnetic resonance imaging (fMRI) of brain function 20 years ago has provided a new methodology for non-invasive measurement of brain function that is now widely used in cognitive neuroscience. Traditionally, fMRI data has been analyzed looking for overall activity changes in brain regions in response to a stimulus or a cognitive task. Now, recent developments have introduced more elaborate, content-based analysis techniques. When multivariate decoding is applied to the detailed patterning of regionally-specific fMRI signals, it can be used to assess the amount of information these encode about specific task-variables. Here we provide an overview over several developments, spanning from applications in cognitive neuroscience (perception, attention, reward, decision making, emotional communication) to methodology (information flow, surface-based searchlight decoding) and medical diagnostics.

In changing environments we need to dynamically re-organize our behavior in order to reach desired goals. This requires strategic switches between different tasks. However, little is known about the neural basis of such adaptation processes. We developed a novel, motivated choice paradigm and used multivariate pattern analysis of fMRI data (MVPA) to investigate this. Subjects chose between three different tasks on a trial-by-trial basis. Task difficulty varied independently for all three tasks, increasing for the chosen task set and decreasing for the non-chosen task sets. This created an exploration-exploitation dilemma for which subjects needed to find a solution. To keep difficulty and error rates low, they had to produce a motivated decision on which task to perform next, based on their estimation of the changing environmental variables. Using MVPA, we were able to predict subjects’ choices from medial PFC, and parietal cortex. Furthermore, we could predict the current difficulty level – the motivational factor driving choices – from reward related areas. Interestingly, the ventro-medial PFC (vmPFC) encoded both the choices and the diffi- culty. This area was previously associated with the processing of stimu- lus-values. Our results suggest that vmPFC is also involved in encoding motivated choices. Further planned analyses will investigate the interaction between areas encoding choices and areas encoding motivational factors in more detail. This should provide important clues on how humans endogenously adapt their behavior to cope with changing environments.

In constantly changing environments, we are forced to dynamically re-organize our behavior in order to reach desired goals. Often, there is no obvious way to adapt to these environments, so different alternative courses of action have to be explored. Once a successful strategy is found, it is exploited until the environment changes and the need for behavioral adaptation arises again. This type of behavior requires endogenous and strategic switches between different behavioral sets. However, the neural implementation of these processes remains unclear. We use multivariate pattern classification of functional MRI data to investigate endogenous adaptations to external task demands in a modified task switching paradigm. More specifically, subjects were free to choose between three different task sets on a trial-by-trial basis. Difficulty varied independently in the three tasks, increasing for the chosen task set and decreasing for the non-chosen task sets. This created an exploration-exploitation dilemma in which subjects had to find the time point at which it is optimal to leave the current task set and switch to a different, possibly easier, one. Strategic choices lead to constantly low difficulty while poor choices lead to high difficulty. Therefore, to keep difficulty low, subjects were forced to constantly monitor the task environment and adapt their behavior accordingly. Behavioral results show that subjects cope efficiently with the task after some training. They track the task environment, keeping difficulty low through strategically placed switches between task sets. Crucially, potential confounds such as task preferences or learning effects do not influence choice behavior. While subjects performed the task we acquired functional MRI data. Using a time resolved decoding approach, we are able to decode whether subjects will stay in the current task or switch away from it from frontopolar cortex. This area is known to be involved in free choice processes. Furthermore, we find predictive information about decisions in posterior parietal cortex, part of a fronto-parietal network associated with cognitive control functions. These results suggest that endogenous behavioral adaptations share a common neural substrate with free choice processes and cognitive control functions - such as the preparation and reconfiguration of task sets. Additional experiments will further illuminate how neural representations of choice-relevant task features (i.e. difficulty) interact with representations of actual choices. This will help us understand how humans strategically adapt their behavior to cope with dynamically changing environments.

We investigated whether a basic deductive inference, namely the modus ponens, is an automatic processing step carried out even in the absence of willingness to reason and awareness of all the premises. We used an exhaustive set of conditional problems concerning numbers, with a subliminal second premise, followed by a target number. Participants evaluated whether the target number was odd or even. The target number could match or not match a valid conclusion endorsed by the premises. We found that evaluations of target numbers matching the conclusion of a modus ponens were faster than the evaluations of numbers following all other types of problems. This finding shows that modus ponens is automatic and can be triggered by subliminal stimuli.

Per lo studio di alcuni compiti linguistici, come la produzione libera di parole di una certa categoria («fluenza semantica»), è necessario disporre di indici quantitativi (come la frequenza, la familiarità, ecc.) per tutti gli elementi della categoria in questione, in quanto non è possibile prevedere quali di essi saranno prodotti dal soggetto sperimentale. La stima preliminare di questi indici su soggetti normali è difficile nel caso di categorie semantiche molto ampie, o in cui siano frequentemente introdotti nuovi esemplari. Per questi motivi, la nostra scelta è caduta sulla categoria «frutta» per la quale presentiamo in questo lavoro i valori di alcune variabili semantico-lessicali raccolti su soggetti normali. La «frutta» è sembrata utile non solo perché ha un numero contenuto di esemplari, ma anche perché essa si articola in sottocategorie, e ciò è vantaggioso per alcuni modelli di applicazione clinica. L’inventario dei membri della categoria è stato fatto mediante un compito preliminare di fluenza semantica, che ha fornito nell’insieme dei soggetti esaminati un elenco di 84 frutti. Un diverso campione di soggetti ha stimato familiarità, prototipicità ed età di acquisizione di ciascuno degli 84 frutti, e la prossimità semantica tra ciascuna delle coppie formate dai 32 frutti più frequentemente prodotti. Sulla matrice di prossimità è stata eseguita una cluster analysis che ha messo in luce in quali sottogruppi la categoria «frutta» possa essere empiricamente suddivisa. Accanto ad una presentazione dei modelli principali del compito di fluenza semantica, si discutono ed esemplificano su due soggetti normali alcune possibili applicazioni cliniche destinate a soggetti cerebrolesi.