| dc.description.abstracteng | Rewards not only shape our value-based choices but also affect our simple perceptual decisions.
Accordingly, behavioral and neural correlates of reward-driven effects on perception have
already been identified at the earliest stages of sensory processing. However, it still needs to be
determined how reward effects arise and are regulated under a multitude of variable contexts
that characterize natural environments. To identify the general principles that govern the
reward-driven modulation of sensory perception, we compared these effects under different
contingencies of reward on performance and sensory modalities of the reward-associated
stimuli (i.e. either visual, auditory, or audiovisual modalities) using behavioral, pupillometry,
and neuroimaging techniques. We hypothesized that whereas some reward effects exhibit
context-dependency, others, such as the dependence on attention and reliance on the long-range
communication of signals across the brain, follow a general, context-independent principle. Our
results in a first behavioral study revealed that whereas performance-contingent reward cues
and previously rewarded stimuli both improved perceptual discrimination, performancecontingent
rewards elicited stronger effects on the response times and pupil dilation compared
to previously rewarded stimuli, indicating a dependency on the mode of reward delivery.
However, no difference between the sensory modalities was found for either performancecontingent
or previously rewarded stimuli. Interestingly, when we tested the previously
rewarded stimuli in a second fMRI study, we found evidence for modality-specific changes in
the effective connectivity between the reward and attention networks and the early visual areas.
Specifically, cross-modal rewards engaged brain areas involved in integrating information
across sensory modalities, namely the Superior Temporal cortex, in addition to those involved
in reward and attentional processing. This finding inspired a third behavioral and fMRI study,
where we tested whether the same principles apply to the contexts where reward cues were
signaled from multiple sensory modalities, i.e., both auditory and visual, and were delivered
continuously after correct responses. Here, we specifically tested whether reward influences the
integration of information across visual and auditory modalities. The results of this last study
revealed overall similar patterns of reward-driven modulations for unisensory and multisensory
stimuli at the behavioral level. Furthermore, although we found evidence for differences
between unisensory and multisensory rewards both in classical reward coding regions and in
higher sensory areas, these differences were overall smaller than the supra-additive threshold
predicted by classical models of multisensory integration. This finding hence indicates that
reward effects, in the setting that we tested, occur at an independent and later stage compared
to multisensory integration, hence affecting the sensory processing in a context-independent
manner. Taken together, this thesis identified key characteristics of reward-associated stimuli
and their mode of delivery that give rise to either dependence or independence from the specific
contexts. Specifically, when rewards were contingent on performance and were delivered
continuously, they mobilized resources to optimize behavior irrespective of the specific sensory
features. However, when rewards were not performance-contingent or were discontinued, they
exhibited modality-specificity, particularly in the way that reward was broadcasted across the
brain. Our findings therefore indicate that the regulation of reward effects on sensory perception
is tightly linked with the optimization of final choices, suggesting that both reward and sensory
systems influence each other to establish adaptive behavior. | de |