Somatosensory and multisensory processing of the body in the human brain

Abstract

Throughout our everyday lives we are exposed to incessant sensory inputs from external sensory stimuli, both proximal and distant, that impact our bodies. Nevertheless the brain constructs a unified representation of the world primarily through continuous integration processing of the simultaneous signals. Two fundamental principles of brain organization were classically associated with the processing and binding of sensory signals. First, according to the sensory division-of-labor principle, multiple sensory inputs are processed independently in different sensory cortices, and then transferred to associative areas where multiple inputs converge and are integrated. The second fundamental principle relates to the topographical organization of the sensory areas, in which nearby points in the stimulated space are represented by the responses of nearby cortical neurons, yielding retinotopic, cochleotopic, and somatosensory or motor somatotopic maps. The prevalent view was that in higher order sensory areas, this topographic organization is gradually replaced by more abstract or complex representations that have much larger receptive fields covering the entire or considerable parts of the sensory world. Converging evidence in the last ten years however has challenged these canonical organization principles to some extent and suggested that the brain might be organized into task-oriented and sensory modality input independent operators rather than sensory-specific areas. Recent studies have shown that topographic organization is a fundamental principle not only in early sensory cortices but also in higher-order processing regions. However these competing views are far from being fully understood or resolved. This thesis focuses on the somatosensory system of humans as a model to study these new views of brain organization using functional MRI (fMRI). Since the somatosensory system is involved in cross-modal integration and exhibits a highly somatotopic organization (at least in the first stages of cortical processing), it has emerged as an appealing model to test these new views and to further postulate a general theoretical framework for such organizational principles. Specifically, I addressed these topics by studying a field which has been less well explored, and tested the topographical biases of the positive and negative blood-oxygenation-level-dependent (BOLD) signals of the tactile evoked imaging responses within and beyond what is commonly assumed to be the somatosensory cortex. Specifically, using continuous and periodic unilateral tactile stimulation of the entire body and applying phase-locked analysis methods, I verified the well-known contralateral topographical organization of the primary somatosensory homunculus (S1), and observed a gradual shift in the representation of different body parts. Furthermore, I showed that the Penfield homunculus should be modified at the medial wall, since the full-body topographic gradient did not reach close to the corpus callosum medially but rather ended at the highest point of the PCG. I further found that different body segments elicit pronounced negative BOLD responses in both the contralateral and ipsilateral hemispheres but show different patterns. In the contralateral hemisphere, S1 contained a sharpening contrast mechanism in which a combination of positive and negative BOLD was evoked for each of the tested body segments, constructing a negative BOLD homunculus. I further showed that negative BOLD also characterizes the ipsilateral cortex, but in contrast to previous studies and to the contralateral hemisphere, the deactivation was not located solely in the ipsilateral homolog of the stimulated body parts but instead was widespread across many parts of the ipsilateral Penfield homunculus. These results, some confirming and elaborating previous observations and some of which are completely novel, suggest a complex pattern of baseline and activity dependent responses in the contralateral and ipsilateral sides, in which negative BOLD responses characterize both primary sensory-motor areas, and suggest that they are an important component of that sharpens the tuning curves of populations of neurons, as was previously reported for topographic gradients in both the visual and the motor cortices. More generally, the results show that a natural tactile stimulus combined with phase-locked analysis provide a powerful and sensitive tool for the mapping of whole-body somatosensory gradients at the group and single-subject levels, and have some clear practical advantages in both clinical and basic research settings. Thus, these methods should be the current gold standard to map whole body gradients. In the second part of the thesis I focused on responses in what is usually considered by most researchers even today as the visual cortex. Here as well I found that, passive touch responses were not confined to the somatosensory cortex, but rather revealed a complex pattern in the visual cortex. I showed that passive touch robustly activated the tool selective parts of the Lateral-Occipital (LO) cortex while almost all other occipital-retinotopic-areas were massively deactivated. These passive touch responses in the visual cortex were specific to the hand and upper trunk stimulation. Resting state functional connectivity analysis showed that although different object selective areas in the lateral and ventral occipital cortex were highly interconnected, only Tactile-LO showed extensive connectivity with parietal areas, and was the sole connecting link of this complex to the hand areas in the primary somatosensory cortex. The results suggest that the LO is a fundamental hub which serves as a node between visual-tool-selectivity and continuous passive (and active) touch information from the hand representation in the primary somatosensory cortex, even in the resting state, probably due to the critical evolutionary role of touch in tool recognition and manipulation. These results might also point to a more general principle that recruitment or deactivation of the visual cortex by other sensory input depends on the ecological relevance of the information conveyed by this input to the task or computation performed by each area or network, and that this task selective recruitment might occur even when the information is received passively without any specific cognitive goal. This is a novel observation that expands the theory to a wider set of conditions and will help formulate better predictions concerning other brain areas. The findings also suggest that this observation might rely on a unique and differential pattern of connectivity for each brain area with the rest of the brain. Altogether the results presented in this thesis suggest that both positive and negative responses reflect topographical biases in which within a given sensory cortex (i.e. somatosensory or visual), each body part only activates functionally relevant subareas, while deactivating the surrounding non-relevant areas. I suggest that the balance between the positive and negative BOLD signal and the ecological relevance of the input at each point in time might be crucial to our understanding of a large variety of intrinsic and extrinsic tasks including low level sensory processing, high-level cognitive functions and multisensory integration.