Neuronal circuits for perceptual inference

Neuroscience in the 21st century is undergoing a remarkable transformation. In the prior century, a common belief was that the brain senses the environment, cognitive processes work with the sensory information that they receive, and then the motor system acts. A new view has emerged that there may actually be very little true sensation and that the brain is constantly making predictions about the world and updating them as sensory information changes. However, when these predictions are inaccurate, weak or unreliable they create abnormal sensations and behaviour, as now implicated in many brain disorders. The basic neuronal circuit that underpins this process of 'perceptual inference' is thought to be the basis for complex behaviour and needs to be better neurobiologically understood in order for clinical teams to improve patient diagnosis and treatment.

An analogy for the brain's neuronal circuit that creates predictions and powerfully influences our behaviour is a busy highway where vehicles moving in different directions and coordinating their movement would reflect the pathways of information flow between neurons in the brain. However, such neural interactions occur at scales that are not visible with the commonly available human brain imaging scanners. Recognising the need for more powerful brain scanners, UK funders, universities and charities have invested substantial amounts (~£150-200M in total) into powerful human scanners and research funding to use them. These scanners are distributed throughout the UK and because they have much higher resolution capabilities than their predecessors, they raise the possibility of being able to visualise some of the brain's information highways with the required sub-millimetre resolution. The UK investment in this domain aims to economically stimulate scientific research and innovation on questions of tremendous societal value and to support UK biosciences advances leading to better scanners, machines, technology and patient treatment.

The problem, returning to the analogy of being able to visualise the highways of information flow in the brain, is that conflicting or surprising results are being obtained with the human scanners. It is currently not clear which aspects of the basic circuit could be visualised in humans. This requires foundational research in a primate model, because in order to truly visualise the flow of traffic in the brain (the brain's vehicles), neurons in the circuit need to be studied and manipulated. This perceptual inference circuit important for complex behaviour involves parts of prefrontal cortex that have evolutionarily differentiated in human and nonhuman primates, thereby requiring macaque monkeys. The primate research can also show which aspects of these circuits could be visualised with powerful brain scanners, such as those available for humans.

We propose to advance the study of neuronal circuits for complex behaviour in a nonhuman primate model and to establish a more direct correspondence to humans by way of using a brain imaging technique that can immediately inform and guide neuroimaging studies with humans. Achieving this will support the delivery on current and future investment in cutting-edge brain imaging systems and technology available for human patients. As information grows on how to interpret what the human brain imaging signal shows, this may in the future also lead to further reduction on the reliance on nonhuman animal research.

This proposal aims to provide an indispensable complement of animal research that would inform theoretical and computational models of the fundamental neuronal circuit, and it could provide crucial neurobiological information on how such circuit functions could be emulated with machines or potentially rehabilitated with non-invasive brain stimulation or brain-machine interfacing devices in patients.

This research can address the timely need for a neuronal circuit model of fronto-temporal feedback and feedforward processes in a primate model, causally tested with optogenetic modulation of pyramidal projection neurons and benchmarked with combined optogenetics and laminar functional Magnetic Resonance Imaging (opto-laminar fMRI) in the primates.

We plan to achieve the three key objectives as follows:

1) Advance a timely neuronal circuit model for perceptual inference in behaving primates: We will use high-density laminar array recordings that can resolve neurons and field potentials across the cortical layers. We will simultaneously record from an ideal auditory fronto-temporal circuit that has the requisite monosynaptic pyramidal neuron interconnections between the areas. This will be conducted as the primates perform efficient auditory and audio-visual predictive sequence learning tasks that will manipulate and allow modelling prediction (including precision) and prediction error variables and signals.

2) Causally test the neuronal circuit model using primate optogenetic neuronal perturbation: We will inject optogenetic viral constructs that express light-sensitive channels exclusively on pyramidal projection neurons between the interconnected fronto-temporal areas, and we will use lasers of different wavelengths to selectively stimulate the neurons. This will allow us to emulate frontal predictive signals onto sensory cortex and test theoretical models on the ideal neuronal oscillatory frequencies that could emulate neuronal information flow in the feedforward and feedback directions.

3) Extend the model from microscopic to mesoscopic scales using sub-millimetre opto-laminar fMRI: We will scan the primates during the tasks using the highest magnetic field MRI scanner in use for awake primates in the UK. The laminar-fMRI results will be benchmarked with the ground truth neuronal circuit data obtained for Aims 1-2.