Publications

Within computational neuroscience, the algorithmic and neural basis of structure learning remains poorly understood. Concept learning is one primary example, which requires both a type of internal model expansion process (adding novel hidden states that explain new observations), and a model reduction process (merging different states into one underlying cause and thus reducing model complexity via meta-learning). Although various algorithmic models of concept learning have been proposed within machine learning and cognitive science, many are limited to various degrees by an inability to generalize, the need for very large amounts of training data, and/or insufficiently established biological plausibility. Using concept learning as an example case, we introduce a novel approach for modeling structure learning-and specifically state-space expansion and reduction-within the active inference framework and its accompanying neural process theory. Our aim is to demonstrate its potential to facilitate a novel line of active inference research in this area. The approach we lay out is based on the idea that a generative model can be equipped with extra (hidden state or cause) "slots" that can be engaged when an agent learns about novel concepts. This can be combined with a Bayesian model reduction process, in which any concept learning-associated with these slots-can be reset in favor of a simpler model with higher model evidence. We use simulations to illustrate this model's ability to add new concepts to its state space (with relatively few observations) and increase the granularity of the concepts it currently possesses. We also simulate the predicted neural basis of these processes. We further show that it can accomplish a simple form of "one-shot" generalization to new stimuli. Although deliberately simple, these simulation results highlight ways in which active inference could offer useful resources in developing neurocomputational models of structure learning. They provide a template for how future active inference research could apply this approach to real-world structure learning problems and assess the added utility it may offer.

The impact of traumatic spinal cord injury on structural integrity, cortical reorganization and ensuing disability is variable and may depend on a dynamic interaction between the severity of local damage and the capacity of the brain for plastic reorganization. We investigated trauma-induced anatomical changes in the spinal cord and brain, and explored their relationship to functional changes in sensorimotor cortex. Structural changes were assessed using cross-sectional cord area, voxel-based morphometry and voxel-based cortical thickness of T1-weighted images in 10 subjects with cervical spinal cord injury and 16 controls. Cortical activation in response to right-sided (i) handgrip; and (ii) median and tibial nerve stimulation were assessed using functional magnetic resonance imaging. Regression analyses explored associations between cord area, grey and white matter volume, cortical activations and thickness, and disability. Subjects with spinal cord injury had impaired upper and lower limb function bilaterally, a 30% reduced cord area, smaller white matter volume in the pyramids and left cerebellar peduncle, and smaller grey matter volume and cortical thinning in the leg area of the primary motor and sensory cortex compared with controls. Functional magnetic resonance imaging revealed increased activation in the left primary motor cortex leg area during handgrip and the left primary sensory cortex face area during median nerve stimulation in subjects with spinal cord injury compared with controls, but no increased activation following tibial nerve stimulation. A smaller cervical cord area was associated with impaired upper limb function and increased activations with handgrip and median nerve stimulation, but reduced activations with tibial nerve stimulation. Increased sensory deficits were associated with increased activations in the left primary sensory cortex face area due to median nerve stimulation. In conclusion, spinal cord injury leads to cord atrophy, cortical atrophy of primary motor and sensory cortex, and cortical reorganization of the sensorimotor system. The degree of cortical reorganization is predicted by spinal atrophy and is associated with significant disability.

Objective The smoothness parameter that characterises the spatial dependence of pixel values in functional brain images is usually estimated empirically from the data. Since this parameter is essential for the assessment of significant changes in brain activity, it is important to know (a) the variance of its estimator and (b) how this variability affects the results of the ensuing statistical analysis. Materials and Methods In this article, we derive an approximate expression for the variance of the smoothness estimator and investigate the effects of this variability on assessing the significance of cerebral activation in statistical parametric maps using a verbal fluency PET activation experiment. Results Our results suggest that, for p values around 0.05, the variability in the p value (due to smoothness estimation) is ∼20%. Conclusion The effect of the assessment of the spatial dependency of the data is far from being negligible, and this suggests a more comprehensive methodology for functional imaging than the one used so far. This work provides a simple tool for taking into account this effect. Index Terms Emission computed tomography, physics and instrumentation—Maps and mapping—Emission computed tomography.

Scale free dynamics are observed in a variety of physical and biological systems. These include neural activity in which evidence for scale freeness has been reported using a range of imaging modalities. Here, we derive the ways in which connections within a network must transform - relative to system size - in order to maintain scale freeness and test these theoretical transformations via simulations. First, we explore the known invariance of planetary motion for orbits varying in size. Using parametric empirical Bayesian modelling and a generic dynamical systems model, we show that we recover Kepler's third law from orbital timeseries, using our proposed transformations; thereby providing construct validation. We then demonstrate that the dynamical critical exponent is inversely proportional to the time rescaling exponent, in the context of coarse graining operations. Using murine calcium imaging data, we then show that the dynamical critical exponent can be estimated in an empirical biological setting. Specifically, we compare dynamical critical exponents - associated with spontaneous and task states in two regions of imaged cortex - that are classified as task-relevant and task-irrelevant. We find, consistently across animals, that the task-irrelevant region exhibits higher dynamical critical exponents during spontaneous activity than during task performance. Conversely, the task-relevant region is associated with higher dynamical critical exponents in task vs. spontaneous states. These data support the idea that higher dynamical critical exponents, within relevant cortical structures, underwrite neuronal processing due to the implicit increase in cross-scale information transmission.

Dynamic causal modelling (DCM) has been applied recently to event-related responses (ERPs) measured with EEG/MEG. DCM attempts to explain ERPs using a network of interacting cortical sources and waveform differences in terms of coupling changes among sources. The aim of this work was to establish the validity of DCM by assessing its reproducibility across subjects. We used an oddball paradigm to elicit mismatch responses. Sources of cortical activity were modelled as equivalent current dipoles, using a biophysical informed spatiotemporal forward model that included connections among neuronal subpopulations in each source. Bayesian inversion provided estimates of changes in coupling among sources and the marginal likelihood of each model. By specifying different connectivity models we were able to evaluate three different hypotheses: differences in the ERPs to rare and frequent events are mediated by changes in forward connections (F-model), backward connections (B-model) or both (FB-model). The results were remarkably consistent over subjects. In all but one subject, the forward model was better than the backward model. This is an important result because these models have the same number of parameters (i.e., the complexity). Furthermore, the FB-model was significantly better than both, in 7 out of 11 subjects. This is another important result because it shows that a more complex model (that can fit the data more accurately) is not necessarily the most likely model. At the group level the FB-model supervened. We discuss these findings in terms of the validity and usefulness of DCM in characterising EEG/MEG data and its ability to model ERPs in a mechanistic fashion.