Oxford Centre for Computational Neuroscience

Professor Edmund T. Rolls

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

Emotion and Decision-Making Explained


The Noisy Brain

Memory, Attention, and Decision-Making

Overview: Rolls has developed a framework for understanding brain mechanisms of emotion and motivation in primates including humans based on mechanisms that represent reward and punishment value in the orbitofrontal cortex but not at earlier stages of cortical processing, and in which the orbitofrontal cortex largely overshadows the amygdala. Rolls and colleagues discovered reward and punishment value, and non-reward, neurons in the primate orbitofrontal cortex; and discovered how taste and olfactory processing in primates occurs to produce reward value representations of the sight, smell, taste and fat texture of food in the orbitofrontal cortex that is important in understanding appetite and obesity. These discoveries are complemented by fMRI  in humans, by investigation of emotional disorders that follow brain damage in humans (641), and by an approach based on these foundations to understanding depression in humans. Key summary descriptions are in B14, B11, B16, 674, 634, 626, 579, and 558. A summary aimed for the general reader in the interests of the public dissemination of science is in B13.

A theory of emotion, motivation, pleasure, and reward; and the principles of their implementation in the brain (B5, B11, B13, B14, B15, B16, 273, 520, 148, 364, 428, 509, 526, 531, 533, 534, 542, 552, 579, 674).

A key principle in primates including humans is that reward value and emotional valence are represented in the orbitofrontal cortex (and amygdala), whereas before that, the representations are about objects and stimuli independently of value, in the inferior temporal visual cortex, the primary taste cortex in the insula, and in the olfactory cortex (B11, B13, B14, 674, 579, B15, B16). This provides for the separation of emotion from perception. The evidence from rodents is different, making them a poor model of emotion and many other cortical systems in primates including humans (B16 Chapter 19).

In this framework, the dopamine neurons are seen as receiving their information from brain regions such as the orbitofrontal cortex, via the ventral striatum and habenula (572, B11, B13, B14, B16, 649). Further, orbitofrontal cortex neurons encode reward value and hence emotion, independently of goal-related actions. The orbitofrontal cortex provides reward-related information to the cingulate cortex for action-outcome learning, and to the basal ganglia for habit-based responses (B11, B13, B14, B16, 579, 606, 649, 674).

A theory of motivation (557, 674).


The discovery of lateral hypothalamic neurons with visual and taste responses to food. These neurons only respond when hunger is present, so encode the reward value of food. (25, 26. 27, 37, 57, B7, B11).


Sensory-specific satiety, discovered by recordings from lateral hypothalamic neurons (46, 47, 55, 57, 59, 68, 69, 82, 89, 104, 234).


The discoveries that taste, olfactory, flavour, and visual sensory-specific satiety are implemented in the primate including human orbitofrontal cortex (124, 216, 285, 334).


The discovery of a basis for brain-stimulation reward: activation of neurons normally activated by natural rewards (12, 48, B1, B7).


The discoveries that the primary taste cortex (119, 120, 437, 552), the primary olfactory cortex (442), and the inferior temporal visual cortex (32) encode information about the identity and intensity of the taste, odour or sight of objects, but not about their reward value (B11, 579, B16).  This is the case for primates and humans, but not for rodents which provide a poor model of processing in these systems of primates including humans (B14, B16).


The discovery of the secondary taste cortex (in the primate orbitofrontal cortex, including lateral and medial parts) (124, 141, 190, 382).


Discovery of the tertiary taste cortex (in the anterior cingulate cortex) (443, 468).


The discovery of multimodal convergence of taste, olfactory and visual inputs onto single neurons in the orbitofrontal cortex to produce flavour (189, 366, 382) using associative learning (211, 212). Age differences in this flavor reward system (544).


The discoveries that the first cortical region in which information about reward value is made explicit in the representation in primates including humans is the orbitofrontal cortex (124, 291, 322, 333, 367, 441, 495, 579, B11, B14, B16), and a second is the anterior cingulate cortex (443, 468, 495, B14, B16).


Discoveries of the roles of sensory-specific satiety, variety in the food available, and top-down cognitive and attentional control reward representations in the orbitofrontal cortex in appetite control and obesity, with implications for the prevention and control of obesity (416, 420, 426, 484, 487, 519, 542, B11, 558, 634, 638).


Discoveries of brain regions where activity is associated with the subjective conscious feeling of pleasure, including the orbitofrontal cortex and anterior cingulate cortex (334, 335, 338, 434, 462, 495, 542, 544, B11, B13, B14, B16).


The discovery that the orbitofrontal cortex is involved in one-trial rule-based ('model-based') visual stimulus-reward learning, and contains negative reward prediction error neurons (79, 337, 579, 627, B11, B13, B14).  The discovery that the lateral orbitofrontal cortex is activated during reward reversal and behavioural inhibition, when behaviour must be changed (337, 575, 627, B13, B14, B16, 626).


The discoveries that oral texture, including viscosity, astringency, and fat texture, and oral temperature, are represented in the primate including human primary taste cortex, the orbitofrontal cortex, and the amygdala (210, 346, 352, 361, 363, 376, 425, 499, 528, 542, 593, 610).


The discovery that fat is sensed by its texture in the mouth (269, 336, 352, 363, 472, 475, 499). The fat sensing is by the coefficient of sliding friction, not by viscosity or by free fatty acids which are separately represented (593).This has important implications for the development of foods with a similar texture, but low energy content (593, 610).


The discovery that the top-down control of reward value and emotion by cognition and attention is implemented by modulation of responsiveness to stimuli in the orbitofrontal cortex and anterior cingulate cortex (381, 434, 437, 440, 442, 480, 488, 520, 530, B11).


The discovery that synergism between the taste of monosodium glutamate and consonant vegetable odours produces the rich delicious flavour umami (158, 243, 279, 330, 417, 469).


The representation of economic value in the orbitofrontal cortex, with different regions responding to monetary gains and losses (288, 424, 623, 627), and both absolute and relative value represented (467) (see 495, B11, B14), with implications for economics (600).


The identification of reward-related decision-making in the ventromedial prefrontal cortex / medial prefrontal cortex area 10, and the representation of confidence about decisions (452, 454, 481, 486, 489, 495, 513, B9, B11, B14, B16).


The discoveries of the principles of operation of the orbitofrontal cortex in humans and other primates (270, 356, 389, 357, 435, 452, 474, 478, 494, 495, 531, 579, 608, B7, B11, B13, B14, B16, 674). 

The connections of the orbitofrontal cortex that help it to implement its functions (190, 608, 620, 649, 665, B16). A recent discovery is of the effective connectivity of the  human orbitofrontal cortex, vmPFC and anterior cingulate cortex, which shows how reward value and emotion can reach the hippocampal memory system to become incorporated in episodic memory (649). This also shows how these cortical regions have connectivity with the septum and basal forebrain cholinergic systems, providing a mechanism that may contribute to the memory impairments produced by vmPFC damage in humans (649, 657), and to the effects of deep brain stimulation in these cortical regions used to treat depression (649, 679).

The discovery that the lateral orbitofrontal cortex has outputs to language systems in the inferior frontal gyrus and provides a route for declarative reports about emotional states (649, 665).

The discovery that, in contrast, the amygdala has relatively little connectivity back to the neocortex in humans, and so may be less involved in consciously experienced emotion in humans than in behavioural and autonomic responses to punishing and rewarding stimuli (665).

The discoveries of face-selective neurons in the amygdala (38, 91, 97), inferior temporal visual cortex (38A, 73, 91, 96, 162), and orbitofrontal cortex (397) (see 412, 451501, 578, B11, B12, B14, B16)


The discoveries of face expression selective neurons in the cortex in the superior temporal sulcus (114, 126, 682) and orbitofrontal cortex (397). Reduced functional and effective connectivity in this region in autism (541, 570, 609).


Impairments in the rapid rule-based reversal of associations between stimuli and reward value in patients with selective lesions of the orbitofrontal cortex and related areas and their relation to emotional changes (188, 203, 331, 354, 641). Also, impairments in impulsivity (353, 362, 394). These discoveries were inspired by the discoveries on neuronal activity in the orbitofrontal cortex, and are relevant to understanding the changes in patients with frontal lobe damage and in patients with borderline personality disorder (B14).

A non-reward attractor theory of depression (559, 572, B13), supported by altered connectivity and activations of the orbitofrontal cortex in depression (564, 579, 583, 588, 591, 592, 596, 623, 626, B14, B16, 679), and a model of the computation of non-reward in the orbitofrontal cortex (562). An introduction to the theory is available as a lecture.

The discoveries that the subjective pleasantness of touch and unpleasantness of pain are represented is the orbitofrontal cortex (322); that the top-down cognitive effects on the pleasantness of touch and the sight of touch are represented in the orbitofrontal and anterior cingulate cortex (440); and that in contrast to the somatosensory cortex, the insula in the ventral somatosensory cortical stream (660) is not activated by the sight of touch, implicating the insula in the representation of touch to one's own body (440, 660).

The roles of the cingulate cortex in emotion, action, and memory, and their disorders, together with the concept that there is no single limbic system (B11, B16, 531, 588, 596, 606, 608, 649, 657, 674). The pregenual anterior cingulate cortex receives reward-related information from the orbitofrontal cortex, and connects to the hippocampal system to provide the value part of episodic memory and the goals for navigation (649, 657). A model for action-outcome learning in which action-related information reaches the posterior cingulate cortex from the parietal cortex, and reaches the supracallosal (dorsal) anterior cingulate cortex (649) from premotor regions as well as posterior cingulate regions, is associated in the supracallosal anterior cingulate cortex with reward outcome information reaching there from the pregenual anterior cingulate cortex and from the orbitofrontal cortex, with outputs connecting to  the midcingulate cortex and the premotor cortex (606, 649, B16). Consistent with this, human sensation-seeking (619), risk-taking (648) and body weight (638) relate to high connectivity between the reward-related medial orbitofrontal cortex and the anterior cingulate cortex.

Basal ganglia: each part of the striatum contains neurons that respond to information received from the cortical areas that project into each striatal region, but this information is brought together by the architecture of the striatum, globus pallidus, and substantia nigra in a way that appears to provide for selection of one behavioural output as a result of competition between mutually inhibitory neurons in these parts of the basal ganglia (80, 84, 147, 174, 181, B7, B11, B16).


Extensions of Granger causality and their application to functional neuroimaging: componential Granger causality (which allows the effects of interactions to be measured); and Granger causality with signal-dependent noise (with J.Feng, T.Ge and colleagues) (505, 530). The use of effective (directed) connectivity in large-scale analyses of the neural bases of depression (583, 679), schizophrenia (602), and autism (609). The application of effective connectivity measured in the human brain to understand brain systems involved in emotion (649, 665); the roles of reward value in emotion and in memory (647, 649, 657); and in somatosensation including the insula identified as part of a ventral somatosensory stream in humans reaching the inferior parietal cortex (660)  with its inputs from the orbitofrontal cortex implicating the insula in affective touch representations (660).

In relation to addiction, the medial orbitofrontal cortex reward system has high functional connectivity in those who tend to drink alcohol and who are sensation-seekers and are impulsive, and the lateral orbitofrontal cortex non-reward system has low functional connectivity in  those who tend to smoke and are impulsive (599). The medial orbitofrontal cortex is also activated by amphetamine (367), and has high functional connectivity in users of arecoline (betel quid) (617).

Reward systems and aesthetics (B10, B11, 509, 532, 556, 574).