Biology subjects

Haynes, J.-D.

Publications and source records attributed to Haynes, J.-D..

4 recordsLinked to original sources

bioRxiv2018

The context-dependent nature of the neural implementation of intentions

Many studies have identified networks in parietal and prefrontal cortex that are involved in intentional action. Yet, knowledge about what these networks exactly encoded is still scarce. In this study we look into the content of those processes. We ask whether the neural representations of intentions are context- and reason-invariant, or whether these processes depend on the context we are in, and the reasons we have for choosing an action. We use a combination of functional magnetic resonance imaging and multivariate decoding to directly assess the context- and reason-dependency of the processes underlying intentional action. We were able to decode action decisions in the same context and for the same reasons from the fMRI data, in line with previous decoding studies. Furthermore, we could decode action decisions across different reasons for choosing an action. Importantly, though, decoding decisions across different contexts was at chance level. These results suggest that for voluntary action, there is considerable context-dependency in intention representations. This suggests that established invariance in neural processes may not reflect an essential feature of a certain process, but that this stable character could be dependent on invariance in the experimental setup, in line with predictions from situated cognition theory.

neuroscience↗

bioRxiv2017

Scale-specific analysis of fMRI data on the irregular cortical surface

To fully characterize the activity patterns on the cerebral cortex as measured with fMRI, the spatial scale of the patterns must be ascertained. Here we address this problem by constructing steerable bandpass filters on the discrete, irregular cortical mesh, using an improved Gaussian smoothing in combination with differential operators of directional derivatives. We demonstrate the utility of the algorithm in two ways. First, using modelling we show that our algorithm yields superior results in numerical precision and spatial uniformity of filter kernels compared to the most widely adopted approach for cortical smoothing. An important interim insight hereby was that the effective scales of information differ from the nominal filter sizes applied to extract them, and thus need to be calculated separately to compare different algorithms on par. Second, we applied the algorithm to an fMRI dataset to assess the scale and pattern form of cortical encoding of information about visual objects in the ventral visual pathway. We found that filtering by our method improved the detection of discriminant information about experimental conditions over previous methods, that the level of categorization (subordinate versus superordinate) of objects was differentially related to the spatial scale of fMRI patterns, and that the spatial scale at which information was encoded increased along the ventral visual pathway. In sum, our results indicate that the proposed algorithm is particularly suited to assess and detect scale-specific information encoding in cortex, and promises further insight into the topography of cortical encoding in the human brain.

neuroscience↗

bioRxiv2017

Switch Independent Task Representations In Frontal And Parietal Cortex

Alternating between two tasks is effortful and impairs performance. Previous functional magnetic resonance imaging (fMRI) studies have found increased activity in fronto-parietal cortex when task switching is required. One possibility is that the additional control demands for switch trials are met by strengthening task representations in the human brain. Alternatively, on switch trials the residual representation of the previous task might impede the buildup of a neural task representation. This would predict weaker task representations on switch trials, thus also explaining the performance costs. To test this, participants were cued to perform one of two similar tasks, with the task being repeated or switched between successive trials. MVPA was used to test which regions encode the tasks and whether this encoding differs between switch and repeat trials. As expected, we found information about task representations in frontal and parietal cortex, but there was no difference in the decoding accuracy of task-related information between switch and repeat trials. Using cross-classification we found that the fronto-parietal cortex encodes tasks using a similar spatial pattern in switch and repeat trials. Thus, task representations in frontal and parietal cortex are largely switch-independent. We found no evidence that neural information about task representations in these regions can explain behavioral costs usually associated with task switching.\n\nSignificance statementAlternating between two tasks is effortful and slows down performance. One possible explanation is that the representations in the human brain need time to build up and are thus weaker on switch trials, explaining performance costs. Alternatively, task representations might even be enhanced in order to overcome the previous task. Here we used a combination of fMRI and a brain classifier to test whether the additional control demands under switching conditions lead to an increased or decreased strength of task representations in fronto-parietal brain regions. We found that task representations are not significantly modulated by switching processes. Thus, task representations in the human brain cannot account for the performance costs associated with alternating between tasks.

neuroscience↗

bioRxiv2016

How to improve parameter estimates in GLM-based fMRI data analysis: cross-validated Bayesian model averaging

In functional magnetic resonance imaging (fMRI), model quality of general linear models (GLMs) for first-level analysis is rarely assessed. In recent work (Soch et al., 2016: \"How to avoid mismodelling in GLM-based fMRI data analysis: cross-validated Bayesian model selection\", NeuroImage, vol. 141, pp. 469-489; DOI: 10.1016/j. neuroimage.2016.07.047), we have introduced cross-validated Bayesian model selection (cvBMS) to infer the best model for a group of subjects and use it to guide second-level analysis. While this is the optimal approach given that the same GLM has to be used for all subjects, there is a much more efficient procedure when model selection only addresses nuisance variables and regressors of interest are included in all candidate models. In this work, we propose cross-validated Bayesian model averaging (cvBMA) to improve parameter estimates for these regressors of interest by combining information from all models using their posterior probabilities. This is particularly useful as different models can lead to different conclusions regarding experimental effects and the most complex model is not necessarily the best choice. We find that cvBMS can prevent not detecting established effects and that cvBMA can be more sensitive to experimental effects than just using even the best model in each subject.

neuroscience↗