- Email: [email protected]
How much can we tell about a person just by looking at the brain?
Abstracts / Neuroscience Research 68S (2010) e4–e52
S3-5-1-3 Cingulate cortex, monitoring, and adjustment Benjamin Yost Hayden , Michael L. Platt Neurobiology, Duke University Making good choices requires monitoring the outcomes of previous choices and incorporating this information into subsequent decisions. Although the cingulate cortex appears to be particularly closely involved in this process, several important questions remain about its precise role. First, the nature of the information monitored by single neurons in dACC remains unclear. To address this question, we recorded responses of single neurons in dACC while monkeys performed a complex eight option choice task. We found that neurons multiplexed information about the location of the chosen target with information about the reward obtained from that target. Furthermore, neurons represented the values of unchosen options that could have been obtained (fictive outcomes) in addition to representing values of chosen options. These results indicate that dACC neurons track information about the world in a richer and more complex manner than previously thought. Another important question concerns the nature of the adjustment signal represented by dACC neurons. To examine this question, we devised a novel choice task that recapitulates the critical assumptions of the patch-leaving problem faced by foragers. We found that dACC neurons track the current likelihood that monkeys will choose to leave a patch. Our data are consistent with the idea that these neurons represent an ongoing scalar representation of the urgency associated with leaving, and suggest that dACC participates in a rise to threshold process. Collectively, these data refine our understanding of the role of cingulate cortex in choices about values. doi:10.1016/j.neures.2010.07.424
e41
The computations of actual choice behaviors, however, go beyond these associations. We highlight two computational capabilities that must be combined with these associations: emulation and extraction of contextual information. We investigated contextual emulation in a social situation, i.e., emulating the other’s value-based decision making. We used human fMRI experiments to examine which brain areas contribute to the emulated-other’s reward prediction error (RPE) to update the subject’s decision variables. Both the ventromedial prefrontal cortex (vmPFC) and temporoparietal junction (TPJ) contributed to the other’s RPE. The TPJ encoded the other’s RPE, but not the self RPE. The vmPFC encoded both the other’s and the self-RPE, but in different regions; therefore, the vmPFC is particularly suited for influencing the subject’s decisions. Thus, the capability to emulate the other can be integrated with the basic form of value-based decision making. We further show, mostly using non-social situations, that information or neural signals beyond current sensory input may greatly improve value-based decision making under the temporal difference learning framework. These signals can enrich input representations for both reward prediction and action selection. Accordingly, they help to achieve better performance in both functions and also make anticipatory preparation and execution possible. These suggest that input constructions by extracting appropriate contextual signals is integral to neural value-based decision-making. This view may resolve some controversial issues in the field, e.g., contextual and salience coding in relation to dopamine activity. doi:10.1016/j.neures.2010.07.426
S3-5-2-1 How much can we tell about a person just by looking at the brain?
S3-5-1-4 Role of basal ganglia for reward-dependent modulation of action
Ryota Kanai
Kae
Our skills and personality differ markedly from one person to another. In this talk, I will give general introduction to the symposium on individual differences as well as the results of my current studies on individual differences and brain structure. I will discuss how individual differences are reflected in localised brain structures and aim to demonstrate with examples how studying neurostructural bases of individual differences illuminates the brain regions underlying perception, cognition and personality trait. In particular, I will show our current results on bistable perception, attentional capture, time perception and numerical cognition. In these studies, we used voxelbased morphometry (VBM) and diffusion tensor imaging (DTI) and found brain structures that reflect individual differences in these perceptual and cognitive skills. Furthermore, we extended our approach to social cognition and personality traits such as impulsivity, distractibility, empathy, extroversion and loneliness. All of these studies imply that individual differences are reflected in localised brain structures and therefore suggest the possibility of predicting individual’s skills and behavioural propensities by simply looking at their brain structure with MRI. Furthermore, we confirmed that at least in some of the cases, the identified structures play causal roles in relevant tasks using neuromodulatory methods such as transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS). Potential applications of these techniques in real world situations will be discussed.
Nakamura 1,2
, Ryuichi
Matsuzaki 1
1
Department of Physiology, Kansai Medical University 2 PRESTO, Japan Science and Technology Agency (JST) It has been hypothesized that the dorsal striatum (DS), once primarily studied in relation to motor function, is also a structure which controls selection of action to obtain rewards. The ventral striatum (VS), on the other hand, has long been studied in relation to reward and motivation. However, the specific nature of VS’s contribution to the motivational behavior is unclear. To test whether different reward information is represented in primate DS and VS, we recorded from single neurons in DS and VS of the same monkeys performing asymmetrically rewarded visually guided and memory-guided saccade tasks. In a given block, one of two target cue positions was associated with a large reward and the other with a small reward. The cue position-reward size association was switched every 16–20 trials. Neurons in the striatum exhibited task related activity during pre-cue, post-cue, saccade, and postreward periods all of which was often modulated by some aspects of reward. We now found these neurons were inhomogeneous in distribution. Neurons showing differential modulation in activity during the pre-cue/movement periods depending on rewarded position, and phasic direction- or rewarddependent modulation in activity during the post-cue and/or peri-saccadic periods, were observed more commonly in DS than in VS. Conversely, neurons showing long-lasting (often >1 s) reward dependent modulation after reward delivery were more prevalent in VS than in DS. In VS, some neurons also exhibited enhancement in activity related to the initiation of the task, switch of the rewarded cue position, or commission of errors. Thus, we propose DS provides spatially relevant reward information which may be useful for executing routine trials while VS provides reward information which may be useful for updating general course of action. Such regional difference may also be related to dissimilar response pattern observed in dopamine and dorsal raphe neurons projecting to DS and VS. doi:10.1016/j.neures.2010.07.425
S3-5-1-5 Computations for value-decision making in nonsocial and social situations: Emulation and extraction of contextual information Hiroyuki Nakahara 1,2 1
Lab for Int Theor Neurosci, RIKEN BSI, Wako, Japan 2 Dpt Comp Intell & Sys Sci, Tokyo Inst of Tech, Yokohama, Japan The computational framework of reinforcement learning or temporal difference learning underlies the recent progress in research on neural value-based decision making and characterizes choice behaviors based on associations between ‘state’ input and ‘action’ output, through reward outcomes or ‘value’.
Insitute of Cognitive Neuroscience, University College London
doi:10.1016/j.neures.2010.07.427
S3-5-2-2 Individual differences in cortico-striatal connections predict control over speed and accuracy in perceptual decision making Birte U. Forstmann Department of Psychology, University of Amsterdam For many everyday life decisions, people face the dilemma that fast decisions tend to be error-prone, whereas accurate decisions tend to be relatively slow. In other words, people can choose to respond more quickly at the cost of making more errors, a phenomenon known as the speed–accuracy tradeoff (SAT). In this talk I discuss the neural substrate of SAT using data from functional magnetic resonance imaging (fMRI), diffusion tensor imaging (DTI), and ultra-high resolution 7 T MRI. In a series of experiments participants were presented cues that indicated different requirements for response speed. Application of the Linear Ballistic Accumulator model (LBA; Brown and Heathcote, 2009) confirmed that these cues selectively affected response thresholds. Individual LBA parameters were then correlated with functional and structural MRI measures, revealing the involvement of the pre-supplementary motor area (pre-SMA) in concert with the striatum; this finding is consistent with the fact that excitatory signals from preSMA to striatum release the motor system from inhibition, thus facilitating
S3-5-1-3 Cingulate cortex, monitoring, and adjustment Benjamin Yost Hayden , Michael L. Platt Neurobiology, Duke University Making good choices requires monitoring the outcomes of previous choices and incorporating this information into subsequent decisions. Although the cingulate cortex appears to be particularly closely involved in this process, several important questions remain about its precise role. First, the nature of the information monitored by single neurons in dACC remains unclear. To address this question, we recorded responses of single neurons in dACC while monkeys performed a complex eight option choice task. We found that neurons multiplexed information about the location of the chosen target with information about the reward obtained from that target. Furthermore, neurons represented the values of unchosen options that could have been obtained (fictive outcomes) in addition to representing values of chosen options. These results indicate that dACC neurons track information about the world in a richer and more complex manner than previously thought. Another important question concerns the nature of the adjustment signal represented by dACC neurons. To examine this question, we devised a novel choice task that recapitulates the critical assumptions of the patch-leaving problem faced by foragers. We found that dACC neurons track the current likelihood that monkeys will choose to leave a patch. Our data are consistent with the idea that these neurons represent an ongoing scalar representation of the urgency associated with leaving, and suggest that dACC participates in a rise to threshold process. Collectively, these data refine our understanding of the role of cingulate cortex in choices about values. doi:10.1016/j.neures.2010.07.424
e41
The computations of actual choice behaviors, however, go beyond these associations. We highlight two computational capabilities that must be combined with these associations: emulation and extraction of contextual information. We investigated contextual emulation in a social situation, i.e., emulating the other’s value-based decision making. We used human fMRI experiments to examine which brain areas contribute to the emulated-other’s reward prediction error (RPE) to update the subject’s decision variables. Both the ventromedial prefrontal cortex (vmPFC) and temporoparietal junction (TPJ) contributed to the other’s RPE. The TPJ encoded the other’s RPE, but not the self RPE. The vmPFC encoded both the other’s and the self-RPE, but in different regions; therefore, the vmPFC is particularly suited for influencing the subject’s decisions. Thus, the capability to emulate the other can be integrated with the basic form of value-based decision making. We further show, mostly using non-social situations, that information or neural signals beyond current sensory input may greatly improve value-based decision making under the temporal difference learning framework. These signals can enrich input representations for both reward prediction and action selection. Accordingly, they help to achieve better performance in both functions and also make anticipatory preparation and execution possible. These suggest that input constructions by extracting appropriate contextual signals is integral to neural value-based decision-making. This view may resolve some controversial issues in the field, e.g., contextual and salience coding in relation to dopamine activity. doi:10.1016/j.neures.2010.07.426
S3-5-2-1 How much can we tell about a person just by looking at the brain?
S3-5-1-4 Role of basal ganglia for reward-dependent modulation of action
Ryota Kanai
Kae
Our skills and personality differ markedly from one person to another. In this talk, I will give general introduction to the symposium on individual differences as well as the results of my current studies on individual differences and brain structure. I will discuss how individual differences are reflected in localised brain structures and aim to demonstrate with examples how studying neurostructural bases of individual differences illuminates the brain regions underlying perception, cognition and personality trait. In particular, I will show our current results on bistable perception, attentional capture, time perception and numerical cognition. In these studies, we used voxelbased morphometry (VBM) and diffusion tensor imaging (DTI) and found brain structures that reflect individual differences in these perceptual and cognitive skills. Furthermore, we extended our approach to social cognition and personality traits such as impulsivity, distractibility, empathy, extroversion and loneliness. All of these studies imply that individual differences are reflected in localised brain structures and therefore suggest the possibility of predicting individual’s skills and behavioural propensities by simply looking at their brain structure with MRI. Furthermore, we confirmed that at least in some of the cases, the identified structures play causal roles in relevant tasks using neuromodulatory methods such as transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS). Potential applications of these techniques in real world situations will be discussed.
Nakamura 1,2
, Ryuichi
Matsuzaki 1
1
Department of Physiology, Kansai Medical University 2 PRESTO, Japan Science and Technology Agency (JST) It has been hypothesized that the dorsal striatum (DS), once primarily studied in relation to motor function, is also a structure which controls selection of action to obtain rewards. The ventral striatum (VS), on the other hand, has long been studied in relation to reward and motivation. However, the specific nature of VS’s contribution to the motivational behavior is unclear. To test whether different reward information is represented in primate DS and VS, we recorded from single neurons in DS and VS of the same monkeys performing asymmetrically rewarded visually guided and memory-guided saccade tasks. In a given block, one of two target cue positions was associated with a large reward and the other with a small reward. The cue position-reward size association was switched every 16–20 trials. Neurons in the striatum exhibited task related activity during pre-cue, post-cue, saccade, and postreward periods all of which was often modulated by some aspects of reward. We now found these neurons were inhomogeneous in distribution. Neurons showing differential modulation in activity during the pre-cue/movement periods depending on rewarded position, and phasic direction- or rewarddependent modulation in activity during the post-cue and/or peri-saccadic periods, were observed more commonly in DS than in VS. Conversely, neurons showing long-lasting (often >1 s) reward dependent modulation after reward delivery were more prevalent in VS than in DS. In VS, some neurons also exhibited enhancement in activity related to the initiation of the task, switch of the rewarded cue position, or commission of errors. Thus, we propose DS provides spatially relevant reward information which may be useful for executing routine trials while VS provides reward information which may be useful for updating general course of action. Such regional difference may also be related to dissimilar response pattern observed in dopamine and dorsal raphe neurons projecting to DS and VS. doi:10.1016/j.neures.2010.07.425
S3-5-1-5 Computations for value-decision making in nonsocial and social situations: Emulation and extraction of contextual information Hiroyuki Nakahara 1,2 1
Lab for Int Theor Neurosci, RIKEN BSI, Wako, Japan 2 Dpt Comp Intell & Sys Sci, Tokyo Inst of Tech, Yokohama, Japan The computational framework of reinforcement learning or temporal difference learning underlies the recent progress in research on neural value-based decision making and characterizes choice behaviors based on associations between ‘state’ input and ‘action’ output, through reward outcomes or ‘value’.
Insitute of Cognitive Neuroscience, University College London
doi:10.1016/j.neures.2010.07.427
S3-5-2-2 Individual differences in cortico-striatal connections predict control over speed and accuracy in perceptual decision making Birte U. Forstmann Department of Psychology, University of Amsterdam For many everyday life decisions, people face the dilemma that fast decisions tend to be error-prone, whereas accurate decisions tend to be relatively slow. In other words, people can choose to respond more quickly at the cost of making more errors, a phenomenon known as the speed–accuracy tradeoff (SAT). In this talk I discuss the neural substrate of SAT using data from functional magnetic resonance imaging (fMRI), diffusion tensor imaging (DTI), and ultra-high resolution 7 T MRI. In a series of experiments participants were presented cues that indicated different requirements for response speed. Application of the Linear Ballistic Accumulator model (LBA; Brown and Heathcote, 2009) confirmed that these cues selectively affected response thresholds. Individual LBA parameters were then correlated with functional and structural MRI measures, revealing the involvement of the pre-supplementary motor area (pre-SMA) in concert with the striatum; this finding is consistent with the fact that excitatory signals from preSMA to striatum release the motor system from inhibition, thus facilitating