Periodic Reporting for period 4 - FREEMIND (FREE the MIND: the neurocognitive determinants of intentional decision)
Reporting period: 2021-09-01 to 2023-02-28
Cognitive flexibility enables decision strategies to be adaptive to environmental and motivational needs. When choosing between similar options with close or no explicit rewards, our decisions are dictated by internal intentions. It remains unclear how the human brain arbitrates between exogenous information and endogenous intentions to enable flexible behaviour. Furthermore, intentional decision, by its nature, is susceptible to changes because of the fluctuation in cognitive states originated from past choices and contexts. We therefore need to investigate these changes in intentional decision under experimental manipulations, by examining objective behavioural and neural measures that do not rely on introspection. FREEMIND aims to establish a multilevel understanding of intentional decision, spanning from neural computations to brain networks to behaviour, through a powerful combination of novel paradigms, cutting-edge brain imaging and innovative methods. Central to the project is formal computational modelling, allowing to establish a quantitative link between data and theory at multiple levels of abstraction. This project will contribute to a mechanistic understanding of intentional behaviour and its changes within and between individuals. In the long term, outcomes of this project would help understand impairments in intentional behaviour in Alzheimer’s and Parkinson’s diseases, which could manifest as apathy, impulsivity and perseveration that affect patients’ quality of life and exacerbate carer burden.
All three subprojects are progressed as planned. Throughout this project, we published 18 journal articles. We have organized a symposium and a training course at international scientific conferences to disseminate our findings and approaches to bridge computational models and brain imaging.
In subproject 1, the research team has developed a set of behavioural paradigms for intentional decision, in which human participants choose between choices with no explicit reward, with probabilistic reward, with hedonic reward or with continuous motor movements. The inclusion of ecological validity and motor complexity in decision-making paradigms enabled a comprehensive research program. We performed rigorous pilots to establish the validity and consistency of the experimental paradigms, and we successfully collected functional MRI and magnetoencephalography (MEG) data while participants were performing the tasks. On computational modelling, the research team has successfully established a platform for whole-brain large-scale simulations for intentional decisions, using biologically realistic neural mass models for individual brain regions.
Subproject 2 considers internal and external factors that may change intentional decisions. In a model-based EEG study, we confirmed that higher reward certainty leads to faster intentional choices between two equally valued options, and humans did not behave randomly in equal choices but established a consistent preference bias towards one option. Using a cognitive model, we demonstrated that the certainty and preference effects related to the speed of evidence accumulation during decision processes. Using multivariate pattern classification, we localized representations of reward certainty and preference choices in electrophysiological recordings. Furthermore, we found specific electrophysiological signatures that tracked the change in accumulation speed predicted by the cognitive model. These results highlighted the neurocognitive signatures of various behavioural effects jointly shaping intentional decisions. In a series of experiments, we have observed that learning shapes the speed and patterns of intentional decisions over consecutive sessions.
Subproject 3 examines individual differences in intentional decisions. We have successfully piloted our behavioural paradigm on an online platform and data collection is now ongoing. On brain imaging, we combined a cognitive model and a biophysical compartment model of diffusion-weighted MRI (DWI) to characterize the neuroanatomical origins of inter-individual variability in the reaction time of simple actions. We have found that motor latency during intentional actions is associated with the DWI measure of neurite density in the bilateral corticospinal tract, indicating a link between inter-individual differences in sensorimotor speed and selective microstructural properties in white matter tracts.
In subproject 1, the research team has developed a set of behavioural paradigms for intentional decision, in which human participants choose between choices with no explicit reward, with probabilistic reward, with hedonic reward or with continuous motor movements. The inclusion of ecological validity and motor complexity in decision-making paradigms enabled a comprehensive research program. We performed rigorous pilots to establish the validity and consistency of the experimental paradigms, and we successfully collected functional MRI and magnetoencephalography (MEG) data while participants were performing the tasks. On computational modelling, the research team has successfully established a platform for whole-brain large-scale simulations for intentional decisions, using biologically realistic neural mass models for individual brain regions.
Subproject 2 considers internal and external factors that may change intentional decisions. In a model-based EEG study, we confirmed that higher reward certainty leads to faster intentional choices between two equally valued options, and humans did not behave randomly in equal choices but established a consistent preference bias towards one option. Using a cognitive model, we demonstrated that the certainty and preference effects related to the speed of evidence accumulation during decision processes. Using multivariate pattern classification, we localized representations of reward certainty and preference choices in electrophysiological recordings. Furthermore, we found specific electrophysiological signatures that tracked the change in accumulation speed predicted by the cognitive model. These results highlighted the neurocognitive signatures of various behavioural effects jointly shaping intentional decisions. In a series of experiments, we have observed that learning shapes the speed and patterns of intentional decisions over consecutive sessions.
Subproject 3 examines individual differences in intentional decisions. We have successfully piloted our behavioural paradigm on an online platform and data collection is now ongoing. On brain imaging, we combined a cognitive model and a biophysical compartment model of diffusion-weighted MRI (DWI) to characterize the neuroanatomical origins of inter-individual variability in the reaction time of simple actions. We have found that motor latency during intentional actions is associated with the DWI measure of neurite density in the bilateral corticospinal tract, indicating a link between inter-individual differences in sensorimotor speed and selective microstructural properties in white matter tracts.
As well as the planned research, we have taken the opportunity to develop new analysis pipelines for multimodal brain imaging. To facilitate future research, we have made our analysis scripts open-source and freely available.
For DWI, the research team has developed an automated method to quantify the changes of microstructural metrics along fiber tracts, using a volume skeletonization algorithm from computational geometry. This approach reduces the noise from individual tractography and generates representative volumetric tract profiles. As a result, our method combined group-level along-tract profiling as well as individual-level fiber tracking, providing a new approach for studying brain, behaviour and cognition relationships across individuals. Since its original publication, this method has been applied to new datasets by the PI’s collaborators, demonstrating its potential impacts.
For MEG, the research team has developed new methods for quantifying the dynamics of MEG oscillatory activity. First, we showed that MEG network oscillatory activity can be accurately described by a pairwise maximum entropy model, from which an energy landscape can be depicted across network states, characterising state probabilities and state transitions from the perspective of attractor dynamics. This approach offers an anatomical-specific and frequency-dependent description of functional dynamics. We are now applying this method to various MEG datasets. Second, we extended microstate analyses, which are conventional for EEG data, to the processing of MEG data. We have published a method paper and an open-source toolbox for the new method.
For DWI, the research team has developed an automated method to quantify the changes of microstructural metrics along fiber tracts, using a volume skeletonization algorithm from computational geometry. This approach reduces the noise from individual tractography and generates representative volumetric tract profiles. As a result, our method combined group-level along-tract profiling as well as individual-level fiber tracking, providing a new approach for studying brain, behaviour and cognition relationships across individuals. Since its original publication, this method has been applied to new datasets by the PI’s collaborators, demonstrating its potential impacts.
For MEG, the research team has developed new methods for quantifying the dynamics of MEG oscillatory activity. First, we showed that MEG network oscillatory activity can be accurately described by a pairwise maximum entropy model, from which an energy landscape can be depicted across network states, characterising state probabilities and state transitions from the perspective of attractor dynamics. This approach offers an anatomical-specific and frequency-dependent description of functional dynamics. We are now applying this method to various MEG datasets. Second, we extended microstate analyses, which are conventional for EEG data, to the processing of MEG data. We have published a method paper and an open-source toolbox for the new method.