Oxford Centre for Computational Neuroscience

Professor Edmund T. Rolls

Discoveries




Cerebral Cortex



Emotion and Decision-Making Explained Cover


The Noisy Brain






This summary is dedicated to all those who have contributed, as indicated by the authors of each paper.

 

Unless otherwise stated, these discoveries apply to humans and other primates. All the investigations are being performed in order to understand better the human brain in health and in disease, and with potential applications to medicine always in mind.

 

The discoveries are being made as part of a program of research to follow sensory processing (including taste, olfactory and visual) through their cortical analysis stages, and then on to brain systems involved in emotion and in memory. By this systematic analysis and comparison between stages and systems it is becoming possible to understand what is being computed at each major stage of cortical processing.

 

This then provides a firm foundation for further investigations into how the processing is performed computationally, to lead to a deep and multidisciplinary understanding of brain function: what is computed, and how it is computed.


These discoveries in turn provide bases for better understanding and treating mental disorders.

Neuroscience of Emotion, reward, pleasure, motivation, decision-making, taste, olfaction, touch, and appetite including implications for the control of food intake and obesity

Key discoveries (see above) include reward value neurons, and non-reward neurons, in the orbitofrontal cortex, leading to a theory of the functions of the orbitofrontal cortex in emotion and depression supported by human neuroimaging and neuropsychological investigations; face and face expression neurons in the orbitofrontal cortex; the secondary taste cortex in the orbitofrontal cortex which represents taste reward value in contrast to the primary taste cortex in the insula which represents taste identity; the representation of fat in food by neuronal activity that encodes the coefficient of sliding friction; olfactory neurons in the orbitofrontal cortex that represent reward value; food reward value neurons; sensory-specific satiety which was discovered by neuronal recordings from food reward value neurons, but which applies to all reward and no punishment systems; the orbitofrontal cortex as a key brain region for brain-stimulation reward; many discoveries with functional neuroimaging and neuropsychology of how these discoveries help to reveal the functions of the orbitofrontal cortex, anterior cingulate cortex, and amygdala in reward processing and emotion in humans; the functions revealed by functional neuroimaging of the human orbitofrontal cortex in the reward value of the sight, smell, taste and oral texture of food, and thereby in human appetite control; a focus for reward value decision-making in the ventromedial prefrontal cortex revealed by fMRI in combination with an integrate-and-fire decision-making model; and a theory of emotion, and of motivation.


Neuroscience of Vision

Key discoveries (see above) in the inferior temporal cortex include face neurons; face expression neurons; invariance for size, position, view, lighting, and spatial frequency of face and object neurons; how information is encoded by sparse distributed representations in the brain at the single neuron and population of neurons levels; that face neurons can learn how to represent new faces in a few trials; that natural scenes greatly reduce the size of face and object neurons' receptive fields, facilitating vision in natural environments; a biologically plausible theory and model (VisNet) of the learning of invariant  face and object neuronal responses in the ventrolateral visual cortical stream; object-based visual motion neurons in the cortex in the superior temporal sulcus and how they are computed; how space in cortical scene areas is represented by allocentric spatial view cells in for example the parahippocampal scene area and hippocampus; the connectivity in humans of a ventromedial visual cortical stream with the theory that spatial view cells are learned by a ventral stream feature combination computation; and a theory of how gain modulation complemented by slow learning enables coordinate transforms in the dorsal visual system to world-based (allocentric) coordinates used for self-motion update of parahippocampal and hippocampal spatial view cells.

Neuroscience of Memory, Spatial Function, and Navigation

Key discoveries (see above) include primate hippocampal and parahippocampal spatial view cells, which revolutionize our understanding of spatial representations for memory and navigation in humans and non-human primates by being so completely different from the rodent hippocampal place cells discovered by O'Keefe and the entorhinal cortex grid cells discovered by the Mosers; the allocentric properties of spatial view cells; the prediction property, related to imagination, of hippocampal spatial cells which respond when a part of the scene is looked at even when the scene is not visible, for example in the dark or when obscured by curtains; episodic memory-like neuronal activity related to one-trial object-location and reward-location memory storage and recall tasks; a neuronal network theory of hippocampal episodic memory and its recall to neocortex, which is the only quantitative theory of these processes; a theory of navigation using spatial view cells responding to landmarks; the spatial view ('Where'), object ('What') and Reward pathways to the human hippocampus; the human ventromedial visual cortical stream with the theory that spatial view cells are learned by a ventral stream feature combination computation; hippocampal whole body motion cells, later found in rodents and termed 'speed cells'; a memory for long-term familiarity by neurons in the perirhinal cortex; a theory of the generation of hippocampal time cells from entorhinal cortex time ramping cells; and the discovery that the human cholinergic basal forebrain neurons receive from the medial orbitofrontal cortex and the theory that this connectivity, and reward inputs to the hippocampus, are important in memory consolidation. For working memory and top-down attention, the discovery of the connectivity of a dorsal prefrontal cortex  system with inferior parietal cortex regions, which complement the dorsolateral prefrontal spatial and inferior prefrontal cortex object systems.

Computational neuroscience theories of brain function and behaviour

Key theories (see above) are of the hippocampus in memory and navigation; of the ventral stream visual pathways for invariant face and object recognition (VisNet); of invariant global and object-based motion recognition in the dorsal visual system; of the functions of the orbitofrontal cortex in reward value, emotion, decision-making, and depression; of how neocortical pyramidal cells can implement the learning of new categories, attractor memory, and top-down recall and attention; of the generation of time cells in the hippocampus; of how alterations in the stability of cortical attractor networks can account for the symptoms of schizophrenia, obsessive compulsive disorder, depression, ADHD, normal aging, and creativity; of the computational utility of diluted connectivity in attractor, pattern association, and competitive networks in the cerebral cortex; of the relation between the mind and the brain utilizing levels of explanation in which causality operates within but not between levels; and of how information in encoded by a sparse distributed firing rate representation in the brain based on my neuronal recordings in many cortical regions. A computational neuroscience approach to understanding brain function is developed in my books Brain Computations and Connectivity (2023, Oxford University Press, Open Access) and Cerebral Cortex: Principles of Operation (Rolls, 2016, Oxford University Press).


Discoveries on the brain bases of mental disorders

Key discoveries (see above) are on depression, in which we find reduced connectivity and activation of the reward-related medial orbitofrontal cortex in depression; and increased connectivity and activation of the non-reward-related lateral orbitofrontal cortex, which supports my theory of depression, and suggests new treatments for depression, for conventional antidepressant drugs do not work on the medial orbitofrontal cortex. Key discoveries on Autism Spectrum Disorder are that the superior temporal cortex face expression and gesture system which we discovered and which is so important in social behaviour has reduced connectivity with the orbitofrontal cortex emotion system. A key discovery on schizophrenia is that there is reduced forward compared to backprojection cortical connectivity, which alters the balance of connectivity and facilitates internally driven instead of externally driven brain activity. A key discovery on normal aging is about how reduced cholinergic activity and reduced connection strengths between neurons can impair the operation of autoassociation networks involved in the storage and recall of memory. In addition, we have discovered that functional and structural differences of the orbitofrontal cortex, anterior cingulate cortex, and hippocampus are associated with lower cognitive performance and with mental health symptoms including depression, and are associated with childhood traumatic events, the family environment, low age of the mother, prolonged nausea and vomiting in pregnancy, and low sleep duration.


Human Cortical Connectivity

Key discoveries (see above) in papers published in 2022-2023 are of a ventromedial visual cortical stream for spatial, 'where', information to reach the human parahippocampal scene area and hippocampal memory and navigation system where spatial view cells are found, which therefore are likely to be built by a feature combination computational process; of the reward and object inputs  in humans to the hippocampal memory system; of the role of the orbitofrontal cortex and pregenual anterior cingulate cortex in providing inputs to the cholinergic basal forebrain system and thereby modulating memory consolidation; of the connectivity of the human dorsal or supracallosal anterior cingulate cortex with premotor and reward systems enabling it to play an important role in action-outcome learning, which is a key function of emotion; of the connectivity of the 'What' and 'Where' parts of the posterior cingulate cortex important in memory; of the connectivity of the different prefrontal cortex systems for different types of working memory; of the human auditory cortical pathways; of the human somatosensory pathways which reach finally the anterior part of the greatly developed human inferior parietal cortex; and of the frontal pole cortex with a theory of how it is involved in exploit vs explore.