The superior parietal role in shifting attention, an interspecies comparison

Directing attention to a spatial location or object improves the detection and discrimination performance of a stimulus presented at that location, even in the absence of eye movements. Spatial attention has been studied in depth in the last decades using invasive recordings of neurons in monkeys and functional magnetic resonance imaging (fMRI) in healthy human subjects and patients. Thanks to the advent of fMRI in macaques in 1998, an essential need became feasible: linking the results of monkey electrophysiology and inactivation studies with human fMRI data. Spatial attention deficits are frequently occurring after stroke. Patients with focal lesions affecting specific brain areas are extremely rare. The first aim of this thesis was therefore to test for the presence of a region in monkey superior parietal lobe (SPL) that might correspond to human SPL, selectively involved in spatial attention shifts. Furthermore, by studying the same task in both species with fMRI, this project lend itself to investigate potential homologies across humans and monkeys, specifically with respect to brain structures recruited by primary networks involved in shifting spatial attention (1), allocating attention to a specific spatial location (2), and finally also default mode brain function (3). All three networks have high clinical relevance, and their breakdown can lead to severe cognitive deficits in humans. By providing results that highlight similarities and differences of these networks across species this thesis established the basis for further causal investigations with direct impact on the interpretation of human data. In the first research chapter (Chapter 3), we investigated the role of monkey superior parietal lobe in the computation of spatial attention shifts across visual hemifields (shift events), compared to moments of allocating attention to one visual hemifield (stay events), during a selective spatial attention task. This was the first study to perform a comparable task in monkeys and humans using fMRI in order to pin down regions implicated in covert spatial attention shifts. Three monkeys were trained to remember and covertly track the relevant stimulus out of two pairs while maintaining a sustained attention baseline. Each pair contained a relevant and an irrelevant stimulus, displayed on the eccentric horizontal meridian in the left and right visual hemifield. Events of interest consisted of a replacement of one stimulus pair by the second stimulus pair. This induced a spatial shift when the relevant stimulus position changed to the opposite visual hemifield (shift event), or no shift when it was replaced by the relevant stimulus of the second pair on the exact same position (stay event). At rare unexpected moments relevant stimulus dimmings (= short flickering) would occur at detection threshold. A fast manual response to the relevant dimming was required to obtain a liquid reward, whereas irrelevant dimmings had to be ignored. When contrasting shift with stay events, we could detect a reliable network activated for spatial attention shifts in monkey. Conversely when contrasting stay left with stay right events, a network responsible for the allocation of contralateral attention was obtained. Whereas contralateral attention modulation occurred mainly in regions where the stimulus was represented, attention shifts recruited a distinct network of areas located more medially in caudal, parietal and prefrontal cortex. Finally, in monkeys, the largest number of shift selective voxels was located in medial parietal cortex, including areas V6, V6A and MIP (medial intraparietal area). This, together with the fact that monkey V6A and MIP, similar to human SPL, integrate visual, somatic and motor information, highlighted this region as a potential homologue of human SPL. In the second research chapter (Chapter 4), we aimed to investigate whether cognitive shifting operations might constitute an underlying functioning principle of the default mode network (DMN) in monkeys. The DMN was first characterized in humans as a large scale cortical network that is consistently deactivated during externally directed cognitive tasks. It is activated during complex internally-directed cognitive processing, but also during passive vigilance of the external environment for unexpected events. We aimed to find the underlying function driving the DMN during these highly different cognitive operations, by using the monkey model. Cognitive shifting operations break down during DMN-associated pathologies; we hence hypothesized that they might constitute an overarching functioning principle of the DMN. In Chapter 4 we tested whether the shift-selective network (of Chapter 3), shared properties with the monkey DMN. A high number of voxels in the monkey shift network overlapped with the monkey DMN, unlike voxels selective for sustained contralateral attention. Raw time courses of core DMN and shift local maxima deactivated for all conditions, which was attenuated during shifting. Finally, high functional correlations across DMN and shift ROIs during rest indicated that these two networks might preferentially interact during task performance. Thus, cognitive shifts (attention shifts) in monkeys recruited DMN regions. Hence, if evolutionary preserved, cognitive shifting may also be one of the basic functions of the human DMN. In order to test this hypothesis in more detail, in Chapter 5, we acquired data in 31 human subjects, and localized their brain networks activated during (1) attention shifts, (2) contralateral allocation of attention, and (3) default mode brain function. Secondly, we investigated the interaction of the human medial shift selective SPL during rest and task performance with the dorsal attention, the ventral attention and the default mode network (DAN, VAN and DMN). To localize the shift and stay networks, we used the same selective attention task as the one performed in monkey (Chapter 3). In humans we detected a similar segregation of shift and stay networks as previously found in monkeys (Chapter 3), and we confirmed the prominent role of human medial SPL during the computation spatial attention shifts. When comparing the human default mode and shift networks, nearly half of the shift voxels overlapped with the DMN, whereas no significant overlap was observed with the contralateral stay network. As in monkey, the raw time courses for both shift and DMN ROIs were negative for all conditions. Unlike in monkeys, DMN ROIs were not shift selective. During functional correlation at rest only the ventral part of human SPL and all other shift-selective regions outside of SPL correlated with DMN-regions. Finally, a separate analysis revealed that human shift-selective dorsal SPL connected differentially to the dorsal, ventral and default mode network based on task demands. Furthermore this SPL region was located at an anatomical position where dorsal, ventral, default mode and frontoparietal control networks (FPCN) converged. Hence, based on task demands, SPL might be able to couple to one of these networks. Shifting in monkeys recruited the DMN, whereas in humans shifting recruited regions at the borders of the FPCN and the DMN. We speculate that the DMN in humans is only recruited during sufficiently demanding cognitive shifts, similar to shifts occurring in everyday cognition. This might be gated by a frontoparietal control network. Overall, this thesis provided the basis for linking spatial attention networks, and default mode brain function across species, enabling further causal investigations of their neural function.nrpages: 240status: publishe