CiteULike: Group: Glimcher_Lab - library [1044 articles]

Free-ranging rhesus monkeys spontaneously individuate and enumerate small numbers of non-solid portions.

Justin N Wood — 2007-04-03T16:00:10-00:00

Cognition (20 March 2007)

Fundamental questions in cognitive science concern the origins and nature of the units that compose visual experience. Here, we investigate the capacity to individuate and store information about non-solid portions, asking in particular whether free-ranging rhesus monkeys (Macaca mulatta) quantify portions of a non-solid substance presented in discrete pouring actions. When presented with portions of carrot pieces poured from a cup into opaque boxes, rhesus picked the box with the greatest number of portions for comparisons of 1 vs. 2, 2 vs. 3, and 3 vs. 4, but not for comparisons of 4 vs. 5 and 3 vs. 6. Additional experiments indicate that rhesus based their decisions on both the number of portions and the total amount of food. These results show that the capacity to individuate non-solid portions is not unique to humans, and does not depend on structures of natural language. Further, the fact that rhesus' ability to represent non-solid portions is constrained by the same 4-item limit typically ascribed to the system of parallel individuation that operates over solid objects suggests that the visual system recruits common working memory processes for retaining information about solid objects and non-solid portions. We discuss our results with respect to theories of visual processing, as well as to the role that the human language faculty may have played in both the evolution and development of quantification.

Differences in Cortical Serotonergic Innervation among Humans, Chimpanzees, and Macaque Monkeys: A Comparative Study

Mary Raghanti — 2007-11-16T13:58:07-00:00

Cereb. Cortex (22 June 2007), bhm089.

In this study, we assess the possibility that the evolution of human intellectual capacities was supported by changes in the supply of serotonin to the frontal cortex. To this end, quantitative comparative analyses were performed among humans, chimpanzees, and macaques. Immunohistochemical methods were used to visualize serotonin transporter-immunoreactive (SERT-ir) axons within the cerebral cortex. Areas 9 and 32 were chosen for evaluation due to their roles in working memory and theory of mind, respectively. Primary motor cortex was also evaluated because it is not associated with higher cognitive functions. The findings revealed that humans do not display a quantitative increase in serotonin innervation. However, the results indicated region- and layer-specific differences among species in serotonergic innervation pattern. Compared with macaques, humans and chimpanzees together displayed a greater density of SERT-ir axons relative to neuron density in layers V/VI. This change was detected in cortical areas 9 and 32, but not in primary motor cortex. Further, morphological specializations, coils of axons, were observed in humans and chimpanzees that were absent in macaques. These features may represent a greater capacity for cortical plasticity exclusive to hominoids. Taken together, these results indicate a significant reorganization of cortical serotonergic transmission in humans and chimpanzees. 10.1093/cercor/bhm089

Dissociation Of working memory from decision making within the human prefrontal cortex.

A Bechara — 2006-03-01T14:48:36-00:00

J Neurosci, Vol. 18, No. 1. (1 January 1998), pp. 428-437.

We tested the hypothesis that cognitive functions related to working memory (assessed with delay tasks) are distinct from those related to decision making (assessed with a gambling task), and that working memory and decision making depend in part on separate anatomical substrates. Normal controls (n = 21), subjects with lesions in the ventromedial (VM) (n = 9) or dorsolateral/high mesial (DL/M) prefrontal cortices (n = 10), performed on (1) modified delay tasks that assess working memory and (2) a gambling task designed to measure decision making. VM subjects with more anterior lesions (n = 4) performed defectively on the gambling but not the delay task. VM subjects with more posterior lesions (n = 5) were impaired on both tasks. Right DL/M subjects were impaired on the delay task but not the gambling task. Left DL/M subjects were not impaired on either task. The findings reveal a cognitive and anatomic double dissociation between deficits in decision making (anterior VM) and working memory (right DL/M). This presents the first direct evidence of such effects in humans using the lesion method and underscores the special importance of the VM prefrontal region in decision making, independent of a direct role in working memory.

The prefrontal cortex and cognitive control.

EK Miller — 2006-01-20T19:33:26-00:00

Nat Rev Neurosci, Vol. 1, No. 1. (October 2000), pp. 59-65.

One of the enduring mysteries of brain function concerns the process of cognitive control. How does complex and seemingly willful behaviour emerge from interactions between millions of neurons? This has long been suspected to depend on the prefrontal cortex--the neocortex at the anterior end of the brain--but now we are beginning to uncover its neural basis. Nearly all intended behaviour is learned and so depends on a cognitive system that can acquire and implement the 'rules of the game' needed to achieve a given goal in a given situation. Studies indicate that the prefrontal cortex is central in this process. It provides an infrastructure for synthesizing a diverse range of information that lays the foundation for the complex forms of behaviour observed in primates.

Representation of action sequence boundaries by macaque prefrontal cortical neurons.

N Fujii — 2008-01-02T16:32:52-00:00

Science, Vol. 301, No. 5637. (29 August 2003), pp. 1246-1249.

Complex biological systems such as human language and the genetic code are characterized by explicit markers at the beginning and end of functional sequences. We report here that macaque prefrontal cortical neurons exhibit phasic peaks of spike activity that occur at the beginning and endpoint of sequential oculomotor saccade performance and have the properties of dynamic start- and end-state encoders accompanying responses to sequential actions. Sequence bounding may thus reflect a general mechanism for encoding biological information.

Tonically active neurons in the striatum encode motivational contexts of action.

M Kimura — 2008-01-02T16:32:00-00:00

Brain Dev, Vol. 25 Suppl 1 (December 2003)

In order to achieve a goal, one procures immediately available rewards, escape from aversive events or endures absence of rewards. The neuronal substrate for these goal-directed actions includes the limbic system and the basal ganglia. In the basal ganglia, classes of projection neurons in the striatum show activity with motivational as well as sensorimotor properties, such as expectation of reward and task schedule for obtaining reward. Tonically active neurons (TANs), presumed cholinergic interneurons in the striatum, respond to reward-associated stimuli, evolve their activity through learning and respond also to aversive event-associated stimuli such as airpuff on the face. A recent study showed that responses to visual cues are less selective to whether the cue instructs reward or no reward. To address this paradox, we asked macaque monkeys to perform a set of visual reaction time tasks while expecting the reward, aversive event or absence of reward. We found that TANs respond to instruction stimuli associated with motivational outcomes but not to unassociated ones, and that they mostly differentiate associated instructions. We also found that the higher percentage of TANs in the caudate nucleus respond to stimuli associated with motivational outcomes than in the putamen, whereas the higher percentage of TANs in the putamen respond to GO signals than in the caudate nucleus especially for an action anticipating a reward. These findings suggest a distinct, pivotal role played by TANs in the caudate nucleus and putamen in encoding instructed motivational contexts for goal-directed action selection and learning in the striatum.

Dissociating uncertainty responses and reinforcement signals in the comparative study of uncertainty monitoring.

JD Smith — 2007-11-25T17:18:54-00:00

J Exp Psychol Gen, Vol. 135, No. 2. (May 2006), pp. 282-297.

Although researchers are exploring animals' capacity for monitoring their states of uncertainty, the use of some paradigms allows the criticism that animals map avoidance responses to error-causing stimuli not because of uncertainty monitored but because of feedback signals and stimulus aversion. The authors addressed this criticism with an uncertainty-monitoring task in which participants completed blocks of trials with feedback deferred so that they could not associate reinforcement signals to particular stimuli or stimulus-response pairs. Humans and 1 of 2 monkeys were able to make cognitive, decisional uncertainty responses that were independent of feedback or reinforcement history within a task. This finding unifies the comparative literature on uncertainty monitoring. The dissociation of performance from reinforcement has theoretical implications, and the deferred-feedback technique has many applications.

Delay of gratification and delay maintenance by rhesus macaques (Macaca mulatta).

TA Evans — 2008-01-02T16:26:43-00:00

J Gen Psychol, Vol. 134, No. 2. (April 2007), pp. 199-216.

The authors tested the self-control of rhesus macaques by assessing if they could refrain from reaching into a food container to maximize the accumulation of sequentially delivered food items (a delay-maintenance task). Three different versions of the task varied the quantity and quality of available food items. In the first 2 versions, food items accumulated across the length of the trial until a monkey consumed the items. In the 3rd task, a single less-preferred food item preceded a single more-preferred food item. Some monkeys delayed gratification even with relatively long delays between deliveries of items. However, the data suggested that self-control, in the majority of tested individuals, was not significantly different across different task versions and that self-control by macaques was not as prevalent in these tasks as it is in chimpanzees and human children.

The uniquely human capacity to throw evolved from a non-throwing primate: an evolutionary dissociation between action and perception.

JN Wood — 2008-01-02T15:56:38-00:00

Biol Lett, Vol. 3, No. 4. (22 August 2007), pp. 360-364.

Humans are uniquely endowed with the ability to engage in accurate, high-momentum throwing. Underlying this ability is a unique morphological adaptation that enables the characteristic rotation of the arm and pelvis. What is unknown is whether the psychological mechanisms that accompany the act of throwing are also uniquely human. Here we explore this problem by asking whether free-ranging rhesus monkeys (Macaca mulatta), which lack both the morphological and neural structures to throw, nonetheless recognize the functional properties of throwing. Rhesus not only understand that human throwing represents a threat, but that some aspects of a throwing event are more relevant than others; specifically, rhesus are sensitive to the kinematics, direction and speed of the rotating arm, the direction of the thrower's eye gaze and the object thrown. These results suggest that the capacity to throw did not coevolve with psychological mechanisms that accompany throwing; rather, this capacity may have built upon pre-existing perceptual processes. These results are consistent with a growing body of work showing that non-human animals often exhibit perceptual competencies that do not show up in their motor responses, suggesting evolutionary dissociations between the systems of perception that provide understanding of the world and those that mediate action on the world.

The evolutionary origins of human patience: temporal preferences in chimpanzees, bonobos, and human adults.

AG Rosati — 2008-01-02T15:54:02-00:00

Curr Biol, Vol. 17, No. 19. (9 October 2007), pp. 1663-1668.

To make adaptive choices, individuals must sometimes exhibit patience, forgoing immediate benefits to acquire more valuable future rewards [1-3]. Although humans account for future consequences when making temporal decisions [4], many animal species wait only a few seconds for delayed benefits [5-10]. Current research thus suggests a phylogenetic gap between patient humans and impulsive, present-oriented animals [9, 11], a distinction with implications for our understanding of economic decision making [12] and the origins of human cooperation [13]. On the basis of a series of experimental results, we reject this conclusion. First, bonobos (Pan paniscus) and chimpanzees (Pan troglodytes) exhibit a degree of patience not seen in other animals tested thus far. Second, humans are less willing to wait for food rewards than are chimpanzees. Third, humans are more willing to wait for monetary rewards than for food, and show the highest degree of patience only in response to decisions about money involving low opportunity costs. These findings suggest that core components of the capacity for future-oriented decisions evolved before the human lineage diverged from apes. Moreover, the different levels of patience that humans exhibit might be driven by fundamental differences in the mechanisms representing biological versus abstract rewards.

Reinforcement Learning: A Survey

Leslie Kaelbling — 2004-12-22T19:38:30-00:00

Journal of Artificial Intelligence Research, Vol. 4 (1996), pp. 237-285.

This paper surveys the field of reinforcement learning from a computer-science perspective. It is written to be accessible to researchers familiar with machine learning. Both the historical basis of the field and a broad selection of currentwork are summarized. Reinforcement learning is the problem faced by an agent that learns behavior through trial-and-error interactions with a dynamic environment. The work described here has a resemblance to work in psychology, but differs considerably in...

The choice axiom after twenty years

Duncan Luce — 2007-09-21T12:34:26-00:00

Journal of Mathematical Psychology, Vol. 15, No. 3. (June 1977), pp. 215-233.

Adding Asymmetrically Dominated Alternatives: Violations of Regularity and the Similarity Hypothesis

Joel Huber — 2007-09-20T21:32:54-00:00

The Journal of Consumer Research, Vol. 9, No. 1. (1982), pp. 90-98.

An asymmetrically dominated alternative is dominated by one item in the set but not by another. Adding such an alternative to a choice set can increase the probability of choosing the item that dominates it. This result points to the inadequacy of many current choice models and suggests product line strategies that might not otherwise be intuitively plausible.

Context-Dependent Preferences

Amos Tversky — 2007-09-20T21:27:00-00:00

Management Science, Vol. 39, No. 10. (1993), pp. 1179-1189.

The standard theory of choice-based on value maximization-associates with each option a real value such that, given an offered set, the decision maker chooses the option with the highest value. Despite its simplicity and intuitive appeal, there is a growing body of data that is inconsistent with this theory. In particular, the relative attractiveness of x compared to y often depends on the presence or absence of a third option z, and the "market share" of an option can actually be increased by enlarging the offered set. We review recent empirical findings that are inconsistent with value maximization, and present a context-dependent model that expresses the value of each option as an additive combination of two components: a contingent weighting process that captures the effect of the background context, and a binary comparison process that describes the effect of the local context. The model accounts for observed violations of the standard theory and provides a framework for analyzing context-dependent preferences.

CHOICE UNDER CONFLICT: The Dynamics of Deferred Decision

Amos Tversky — 2007-09-20T21:21:55-00:00

Psychological Science, Vol. 3, No. 6. (1992), pp. 358-361.

-Choice often produces conflict. This notion, however, plays no role in classical decision theory, in which each alternative is assigned a value, and the decision maker selects from every choice set the option with the highest value. We contrast this principle of value maximization with the hypothesis that the option to delay choice or seek new alternatives is more likely to be selected when conflict is high than when it is low. This hypothesis is supported by several studies showing that the tendency to defer decision, search for new alternatives, or choose the default option can be in-creased when the offered set is enlarged or improved, contrary to the principle of value maximization.

Prospect relativity: how choice options influence decision under risk.

N Stewart — 2007-09-20T19:50:47-00:00

J Exp Psychol Gen, Vol. 132, No. 1. (March 2003), pp. 23-46.

In many theories of decision under risk (e.g., expected utility theory, rank-dependent utility theory, and prospect theory), the utility of a prospect is independent of other options in the choice set. The experiments presented here show a large effect of the available options, suggesting instead that prospects are valued relative to one another. The judged certainty equivalent for a prospect is strongly influenced by the options available. Similarly, the selection of a preferred prospect is strongly influenced by the prospects available. Alternative theories of decision under risk (e.g., the stochastic difference model, multialternative decision field theory, and range frequency theory), where prospects are valued relative to one another, can provide an account of these context effects.

Discrete choice models

Michael Bierlaire — 2007-09-20T15:00:37-00:00

(1998), pp. 203-227.

Discrete choice models have played an important role in transportation modeling for the last 25 years. They are namely used to provide a detailed representation of the complex aspects of transportation demand, based on strong theoretical justifications. Moreover, several packages and tools are available to help practionners using these models for real applications, making discrete choice models more and more popular. Discrete choice models are powerful but complex. The art of finding the appropriate model for a particular application requires from the analyst both a close familiarity with the reality under interest and a strong understanding of the methodological and theoretical background of the model. The main theoretical aspects of discrete choice models are reviewed in this paper. The main assumptions used to derive discrete choice models in general, and random utility models in particular, are covered in detail. The Multinomial Logit Model, the Nested Logit Model and the Generalized Extreme Value model are also discussed.

Generalized random utility model

Joan Walker — 2007-09-19T18:16:35-00:00

Mathematical Social Sciences, Vol. 43, No. 3. (July 2002), pp. 303-343.

Researchers have long been focused on enriching Random Utility Models (RUMs) for a variety of reasons, including to better understand behavior, to improve the accuracy of forecasts, and to test the validity of simpler model structures. While numerous useful enhancements exist, they tend to be discussed and applied independently from one another. This paper presents a practical, generalized model that integrates many enhancements that have been made to RUM. In the generalized model, RUM forms the core, and then extensions are added that relax simplifying assumptions and enrich the capabilities of the basic model. The extensions that are included are: - Flexible Disturbances in order to allow for a rich covariance structure and enable estimation of unobserved heterogeneity through, for example, random parameters; - Latent Variables in order to provide a richer explanation of behavior by explicitly representing the formation and effects of latent constructs such as attitudes and perceptions; - Latent Classes in order to capture latent segmentation in terms of, for example, taste parameters, choice sets, and decision protocols; and - Combining Revealed Preferences and Stated Preferences in order to draw on the advantages of the two types of data, thereby reducing bias and improving efficiency of the parameter estimates. The paper presents a unified framework that encompasses all models, describes each enhancement, and shows relationships between models including how they can be integrated. These models often result in functional forms composed of complex multidimensional integrals. Therefore, an estimation method consisting of Simulated Maximum Likelihood Estimation with a kernel smooth simulator is reviewed, which provides for practical estimation. Finally, the practicality and usefulness of the generalized model and estimation technique is demonstrated by applying it to a case study.

Stochastic expected utility theory

Blavatskyy — 2007-07-12T05:00:21-00:00

Journal of Risk and Uncertainty, Vol. 34, No. 3. (June 2007), pp. 259-286.

Incorporating a stochastic element into decision theories

Graham Loomes — 2007-09-19T17:55:38-00:00

European Economic Review, Vol. 39, No. 3-4. (April 1995), pp. 641-648.

Recent papers by Harless and Camerer (1994) and Hey and Orme (1994) have employed two rather different models of errors in decision making under uncertainty. The present paper develops a third approach and shows how different stochastic specifications of the same deterministic `core' theory may generate very different (and sometimes surprising) hypotheses.

Stochastic Choice Under Risk

Pavlo Blavatskyy — 2007-09-19T17:48:50-00:00

An individual makes random errors when evaluating the expected utility of a risky lottery. Errors are symmetrically distributed around zero as long as an individual does not make transparent mistakes such as choosing a risky lottery over its highest possible outcome for certain. This stochastic decision theory explains many well-known violations of expected utility theory such as the fourfold pattern of risk attitudes, the discrepancy between certainty equivalent and probability equivalent elicitation methods, the preference reversal phenomenon, the generalized common consequence effect (the Allais paradox), the common ratio effect and the violations of the betweenness.

The debate over dopamines role in reward: the case for incentive salience

Berridge — 2007-03-31T17:58:38-00:00

Psychopharmacology, Vol. 191, No. 3. (April 2007), pp. 391-431.

The short-latency dopamine signal: a role in discovering novel actions?

Peter Redgrave — 2006-11-20T22:53:28-00:00

Nature Reviews Neuroscience, Vol. 7, No. 12. (08 November 2006), pp. 967-975.

Reward Prediction Error Computation in the Pedunculopontine Tegmental Nucleus Neurons

Yasushi Kobayashi — 2007-06-29T02:35:43-00:00

Annals of the New York Academy of Sciences, Vol. 1104, No. 1. (May 2007), pp. 310-323.

Substantia nigra/ventral tegmental reward prediction error disruption in psychosis

GK Murray — 2007-08-09T00:41:02-00:00

Molecular Psychiatry, Vol. aop, No. current.

Height as a potential indicator of early life events predicting Parkinson's disease: A case-control study.

Paolo Ragonese — 2007-09-18T13:56:48-00:00

Mov Disord (12 September 2007)

Aim of this study was to investigate the relationship between height in young adult age and Parkinson's disease (PD) risk. We included 266 persons affected by idiopathic PD. Patients were matched by age and sex to 266 controls by a random selection from the municipality of residence. We collected information about height preceding PD from official documents where these characteristics referred to young adult age (nearly 30 years). We compared height in cases and controls by calculating differences in mean distribution and by chi(2) analyses. Crude and adjusted odds ratios (OR) and 95% confidence intervals (CI) were calculated by logistic regression models. Mean height was significantly lower in persons affected by PD compared to controls (P = 0.03). Difference was significant only in men (P = 0.001). Logistic regression models showed an inverse association between height and PD (OR 0.35; CI 0.16, 0.79; P < 0.01 comparing individuals in the highest percentiles of height with those in the lowest). Our results indicate an association between height and PD in men. Considering that dopamine sensitivity in the hypothalamic-pituitary axis is related to adult height, our findings suggest a relationship between PD and factors modulating somatic growth early in life. (c) 2007 Movement Disorder Society.

Modeling Methods for Discrete Choice Analysis

Moshe Ben-Akiva — 2007-09-12T23:27:53-00:00

Marketing Letters, Vol. 8, No. 3. (1 July 1997), pp. 273-286.

This paper introduces new forms, sampling and estimation approaches fordiscrete choice models. The new models include behavioral specifications oflatent class choice models, multinomial probit, hybrid logit, andnon-parametric methods. Recent contributions also include new specializedchoice based sample designs that permit greater efficiency in datacollection. Finally, the paper describes recent developments in the use ofsimulation methods for model estimation. These developments are designed toallow the applications of discrete choice models to a wider variety ofdiscrete choice problems.

Area under the curve as a measure of discounting.

J Myerson — 2007-09-12T15:39:45-00:00

J Exp Anal Behav, Vol. 76, No. 2. (September 2001), pp. 235-243.

We describe a novel approach to the measurement of discounting based on calculating the area under the empirical discounting function. This approach avoids some of the problems associated with measures based on estimates of the parameters of theoretical discounting functions. The area measure may be easily calculated for both individual and group data collected using any of a variety of current delay and probability discounting procedures. The present approach is not intended as a substitute for theoretical discounting models. It is useful, however, to have a simple, univariate measure of discounting that is not tied to any specific theoretical framework.

Three-month stability of delay and probability discounting measures.

Y Ohmura — 2007-09-07T21:22:15-00:00

Exp Clin Psychopharmacol, Vol. 14, No. 3. (August 2006), pp. 318-328.

Psychopharmacologists are interested in delay and probability discounting because the tendency to discount the value of future and uncertain rewards has been linked with drug dependency. However, relatively little is known about the long-term stability of discounting measures typically studied in clinical psychopharmacology. To evaluate the stability of discounting over a 3-month period, the authors compared points of subjective equality (indifference points) with those collected from the same subjects 3 months earlier. Seven delay periods, ranging from 1 week to 25 years, and 7 probability values, ranging from .95 to .05, were assessed in an undergraduate sample (n=22, delay discounting; n=18, probability discounting). The authors examined both differential stability (stability of individual differences) and absolute stability (stability of the group mean) of delay and probability discounting measures as well as their respective indifference points. The results demonstrate that standard delay and probability discounting parameters (e.g., hyperbolic k and area under the curve) had both differential stability and absolute stability across 3 months. Moreover, most indifference points in the delay and probability discounting tasks demonstrated both differential and absolute stability. All together, these results suggest that delay and probability parameters are stable enough to predict future behavior, such as substance abuse. Additional findings indicated that a hyperbolic function fitted the data better than an exponential function and that delay and probability discounting parameters were not significantly correlated.

Lateral habenula as a source of negative reward signals in dopamine neurons

Masayuki Matsumoto — 2007-05-24T01:27:28-00:00

Nature (23 May 2007)

Escaping the tyranny of choice: when fewer attributes make choice easier

Barbara Fasolo — 2007-09-05T15:10:50-00:00

Marketing Theory, Vol. 7, No. 1. (1 March 2007), pp. 13-26.

In the age of the Internet and easy access to almost infinite information, the problem of information overload among consumers is bound to become of great importance to marketers. By means of simulations we show that this tyranny of choice' is avoidable. Consumers can neglect most product information and yet make good choices, so long as either there is no conflict among the product attributes or the attributes are unequally important. In these conditions, only one attribute is enough to select a good option - one within ten percent of the highest value possible. We conclude with marketing implications of these findings. 10.1177/1470593107073842

Search goal tunes visual features optimally.

V Navalpakkam — 2007-03-12T14:36:27-00:00

Neuron, Vol. 53, No. 4. (15 February 2007), pp. 605-617.

How does a visual search goal modulate the activity of neurons encoding different visual features (e.g., color, direction of motion)? Previous research suggests that goal-driven attention enhances the gain of neurons representing the target's visual features. Here, we present mathematical and behavioral evidence that this strategy is suboptimal and that humans do not deploy it. We formally derive the optimal feature gain modulation theory, which combines information from both the target and distracting clutter to maximize the relative salience of the target. We qualitatively validate the theory against existing electrophysiological and psychophysical literature. A surprising prediction is that it is sometimes optimal to enhance nontarget features. We provide experimental evidence toward this through psychophysics experiments on human subjects, thus suggesting that humans deploy the optimal gain modulation strategy.

Statistics of Midbrain Dopamine Neuron Spike Trains in the Awake Primate

Hannah Bayer — 2007-07-19T11:11:14-00:00

J Neurophysiol (5 July 2007), 01140.2006.

Work in behaving primates indicates that midbrain dopamine neurons encode a prediction error, the difference between an obtained reward and the reward expected. Studies of dopamine action potential timing in the alert and anaesthetized rat indicate that dopamine neurons respond in tonic and phasic modes, a distinction that has been less well characterized in the primates. We used spike train models to examine the relationship between the tonic and burst modes of activity in dopamine neurons while monkeys were performing a reinforced visuo-saccadic movement task. We studied spiking activity during four task-related intervals; two of these were intervals during which no task-related events occurred, while two were periods marked by task-related phasic activity. We found that dopamine neuron spike trains during the intervals when no events occurred were well described as tonic. Action potentials appeared to be independent, to occur at low frequency, and to be almost equally well described by Gaussian and Poisson-like (Gamma) processes. Unlike in the rat, interspike intervals as low as 20 ms were often observed during these presumptively tonic epochs. Having identified these periods of presumptively tonic activity we were able to quantitatively define phasic modulations (both increases and decreases in activity) during the intervals in which task-related events occurred. This analysis revealed that the phasic modulations of these neurons include both bursting, as has been described previously, and pausing. Together bursts and pauses seemed to provide a continuous, although non-linear, representation of the theoretically defined reward prediction error of reinforcement learning. 10.1152/jn.01140.2006

Monotonic coding of numerosity in macaque lateral intraparietal area.

JD Roitman — 2007-08-21T16:20:34-00:00

PLoS Biol, Vol. 5, No. 8. (August 2007)

As any child knows, the first step in counting is summing up individual elements, yet the brain mechanisms responsible for this process remain obscure. Here we show, for the first time, that a population of neurons in the lateral intraparietal area of monkeys encodes the total number of elements within their classical receptive fields in a graded fashion, across a wide range of numerical values (2-32). Moreover, modulation of neuronal activity by visual quantity developed rapidly, within 100 ms of stimulus onset, and was independent of attention, reward expectations, or stimulus attributes such as size, density, or color. The responses of these neurons resemble the outputs of "accumulator neurons" postulated in computational models of number processing. Numerical accumulator neurons may provide inputs to neurons encoding specific cardinal values, such as "4," that have been described in previous work. Our findings may explain the frequent association of visuospatial and numerical deficits following damage to parietal cortex in humans.

High-Speed Imaging Reveals Neurophysiological Links to Behavior in an Animal Model of Depression

Raag Airan — 2007-07-06T15:29:21-00:00

Science (5 July 2007), 1144400.

The hippocampus is one of several brain areas thought to play a central role in affective behaviors, but the underlying local network dynamics are not understood. We used quantitative voltage-sensitive dye imaging to probe hippocampal dynamics with millisecond resolution in brain slices following bidirectional modulation of affective state in rodent models of depression. We found that a simple measure of real-time activity, stimulus-evoked percolation of activity through the dentate gyrus relative to the hippocampal output subfield, accounted for induced changes in animal behavior independent of the underlying mechanism of action of the treatments. Our results define a network-level neurophysiological endophenotype for affective behavior and suggest an approach to understanding circuit-level substrates underlying psychiatric disease symptoms. 10.1126/science.1144400

AIC model selection using Akaike weights.

EJ Wagenmakers — 2007-08-08T21:15:51-00:00

Psychon Bull Rev, Vol. 11, No. 1. (February 2004), pp. 192-196.

The Akaike information criterion (AIC; Akaike, 1973) is a popular method for comparing the adequacy of multiple, possibly nonnested models. Current practice in cognitive psychology is to accept a single model on the basis of only the "raw" AIC values, making it difficult to unambiguously interpret the observed AIC differences in terms of a continuous measure such as probability. Here we demonstrate that AIC values can be easily transformed to so-called Akaike weights (e.g., Akaike, 1978, 1979; Bozdogan, 1987; Burnham & Anderson, 2002), which can be directly interpreted as conditional probabilities for each model. We show by example how these Akaike weights can greatly facilitate the interpretation of the results of AIC model comparison procedures.

Model Comparisons and R$^2$

Richard Anderson-Sprecher — 2006-09-25T15:56:48-00:00

The American Statistician, Vol. 48, No. 2. (May 1994), pp. 113-117.

Much of the confusion surrounding interpretation and application of the coefficient of determination, R$^2$, can be alleviated if it is defined explicitly as a comparison of a given model to the null model EY = $β_0$. The model-comparison definition allows R$^2$ to be easily generalized, and standard extensions such as coefficients of partial determination are seen to be special cases of this generalization. Formulas become simpler, more unified, and more easily understood. Commonly cited problem areas such as R$^2$ for the no-intercept model and model comparisons using different values of R$^2$ are also clarified by this perspective.

Likelihood Ratio Tests for Model Selection and Non-Nested Hypotheses

Quang Vuong — 2005-03-17T13:57:55-00:00

Econometrica, Vol. 57, No. 2. (1989), pp. 307-333.

In this paper, we develop a classical approach to model selection. Using the Kullback-Leibler Information Criterion to measure the closeness of a model to the truth, we propose simple likelihood-ratio based statistics for testing the null hypothesis that the competing models are equally close to the true data generating process against the alternative hypothesis that one model is closer. The tests are directional and are derived successively for the cases where the competing models are non-nested, overlapping, or nested and whether both, one, or neither is misspecified. As a prerequisite, we fully characterize the asymptotic distribution of the likelihood ratio statistic under the most general conditions. We show that it is a weighted sum of chi-square distribution or a normal distribution depending on whether the distributions in the competing models closest to the truth are observationally identical. We also propose a test of this latter condition.

How Not to Lie with Statistics: Avoiding Common Mistakes in Quantitative Political Science

Gary King — 2006-08-28T16:22:01-00:00

American Journal of Political Science, Vol. 30, No. 3. (1986), pp. 666-687.

This article identifies a set of serious theoretical mistakes appearing with troublingly high frequency throughout the quantitative political science literature. These mistakes are all based on faulty statistical theory or on erroneous statistical analysis. Through algebraic and interpretive proofs, some of the most commonly made mistakes are explicated and illustrated. The theoretical problem underlying each is highlighted, and suggested solutions are provided throughout. It is argued that closer attention to these problems and solutions will result in more reliable quantitative analyses and more useful theoretical contributions

Behavioural improvements with thalamic stimulation after severe traumatic brain injury

ND Schiff — 2007-08-01T20:22:16-00:00

Nature, Vol. 448, No. 7153. (2007), pp. 600-603.

Neuro-Dynamic Programming (Optimization and Neural Computation Series, 3)

Dimitri Bertsekas — 2005-05-20T19:31:27-00:00

(01 May 1996)

Inactivation of dorsolateral striatum enhances sensitivity to changes in the action-outcome contingency in instrumental conditioning.

HH Yin — 2006-11-27T18:40:37-00:00

Behav Brain Res, Vol. 166, No. 2. (30 January 2006), pp. 189-196.

Actions become compulsive when they are no longer controlled by their consequences. Compulsivity can be assessed using the omission procedure in which animals are required to withhold a previously reinforced action to earn reward. The current study tested the hypothesis that inactivation of the dorsolateral striatum (DLS), a structure implicated in habitual behavior, can enhance sensitivity to changes in the action-outcome contingency during omission training, thus leading to a reduction in compulsive responding. Over 10 days rats were trained to press a freely available lever for sucrose reward delivered on interval schedules of reinforcement. After learning to press the lever at a stable and high rate, rats in the omission group received a session in which the rewards were now delayed by pressing the lever; i.e. withholding lever pressing resulted in increased access to reward. A control group was yoked to the omission group and received the same number and pattern of reward delivery but without the omission contingency. Half the rats in each group received infusions of vehicle into the DLS prior to this training whereas the remainder received an infusion of the GABA-A receptor agonist muscimol. On the next day, the effect of these treatments was assessed on a probe test in which the tendency of the various groups to press the lever was assessed in extinction and without drug infusion. Rats that received vehicle infusions prior to the omission session showed complete insensitivity to the newly imposed omission contingency. In contrast, rats given the infusion of muscimol selectively reduced lever pressing compared to yoked controls. Thus, extended training with interval schedules resulted in compulsive lever pressing that prevented the learning of the omission contingency, whereas inactivation of the DLS appeared to enhance the rats' sensitivity to this change in the action-outcome contingency.

Reward prediction in primate basal ganglia and frontal cortex

Wolfram Schultz — 2007-07-23T18:32:24-00:00

Neuropharmacology, Vol. 37, No. 4-5. (5 April 1998), pp. 421-429.

Reward information is processed in a limited number of brain structures, including fronto-basal ganglia systems. Dopamine neurons respond phasically to primary rewards and reward-predicting stimuli depending on reward unpredictability but without discriminating between rewards. These responses reflect `errors' in the prediction of rewards in correspondence to learning theories and thus may constitute teaching signals for appetitive learning. Neurons in the striatum (caudate, putamen, ventral striatum) code reward predictions in a different manner. They are activated during several seconds when animals expect predicted rewards. During learning, these activations occur initially in rewarded and unrewarded trials and become subsequently restricted to rewarded trials. This occurs in parallel with the adaptation of reward expectations by the animals, as inferred from their behavioral reactions. Neurons in orbitofrontal cortex respond differentially to stimuli predicting different liquid rewards, without coding spatial or visual features. Thus, different structures process reward information processed in different ways. Whereas dopamine neurons emit a reward teaching signal without indicating the specific reward, striatal neurons adapt expectation activity to new reward situations, and orbitofrontal neurons process the specific nature of rewards. These reward signals need to cooperate in order for reward information to be used for learning and maintaining approach behavior.

Goal-directed instrumental action: contingency and incentive learning and their cortical substrates

Bernard Balleine — 2007-07-23T18:29:47-00:00

Neuropharmacology, Vol. 37, No. 4-5. (5 April 1998), pp. 407-419.

Instrumental behaviour is controlled by two systems: a stimulus-response habit mechanism and a goal-directed process that involves two forms of learning. The first is learning about the instrumental contingency between the response and reward, whereas the second consists of the acquisition of incentive value by the reward. Evidence for contingency learning comes from studies of reward devaluation and from demonstrations that instrumental performance is sensitive not only the probability of contiguous reward but also to the probability of unpaired rewards. The process of incentive learning is evident in the acquisition of control over performance by primary motivational states. Preliminary lesion studies of the rat suggest that the prelimibic area of prefrontal cortex plays a role in the contingency learning, whereas the incentive learning for food rewards involves the insular cortex.

Neural signature of fictive learning signals in a sequential investment task

Terry Lohrenz — 2007-07-23T13:26:29-00:00

PNAS, Vol. 104, No. 22. (29 May 2007), pp. 9493-9498.

Reinforcement learning models now provide principled guides for a wide range of reward learning experiments in animals and humans. One key learning (error) signal in these models is experiential and reports ongoing temporal differences between expected and experienced reward. However, these same abstract learning models also accommodate the existence of another class of learning signal that takes the form of a fictive error encoding ongoing differences between experienced returns and returns that "could-have-been-experienced" if decisions had been different. These observations suggest the hypothesis that, for all real-world learning tasks, one should expect the presence of both experiential and fictive learning signals. Motivated by this possibility, we used a sequential investment game and fMRI to probe ongoing brain responses to both experiential and fictive learning signals generated throughout the game. Using a large cohort of subjects (n = 54), we report that fictive learning signals strongly predict changes in subjects' investment behavior and correlate with fMRI signals measured in dopaminoceptive structures known to be involved in valuation and choice. 10.1073/pnas.0608842104

Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians.

J Carpenter — 2007-07-20T21:23:47-00:00

Stat Med, Vol. 19, No. 9. (15 May 2000), pp. 1141-1164.

Since the early 1980s, a bewildering array of methods for constructing bootstrap confidence intervals have been proposed. In this article, we address the following questions. First, when should bootstrap confidence intervals be used. Secondly, which method should be chosen, and thirdly, how should it be implemented. In order to do this, we review the common algorithms for resampling and methods for constructing bootstrap confidence intervals, together with some less well known ones, highlighting their strengths and weaknesses. We then present a simulation study, a flow chart for choosing an appropriate method and a survival analysis example.

Decisions, decisions, decisions: choosing a biological science of choice.

P Glimcher — 2007-07-20T18:44:53-00:00

Neuron, Vol. 36, No. 2. (10 October 2002), pp. 323-332.

Behavioral ecologists argue that evolution drives animal behavior to efficiently solve the problems animals face in their environmental niches. The ultimate evolutionary causes of decision making, they contend, can be found in economic analyses of organisms and their environments. Neurobiologists interested in how animals make decisions have, in contrast, focused their efforts on understanding the neurobiological hardware that serves as a more proximal cause of that same behavior. Describing the flow of information within the nervous system without regard to these larger goals has been their focus. Recent work in a number of laboratories has begun to suggest that these two approaches are beginning to fuse. It may soon be possible to view the nervous system as a representational process that solves the mathematically defined economic problems animals face by making efficient decisions. These developments in the neurobiological theory of choice, and the new schema they imply, form the subject of this article.

Probabilistic reasoning by neurons

Tianming Yang — 2007-06-04T03:11:35-00:00

Nature (03 June 2007)

Translational principles of deep brain stimulation

Morten Kringelbach — 2007-07-20T18:29:33-00:00

Nat Rev Neurosci, Vol. 8, No. 8. (2007), pp. 623-635.

Temporal Patterning of Saccadic Eye Movement Signals

Daniel Kimmel — 2007-07-19T13:42:35-00:00

J. Neurosci., Vol. 27, No. 29. (18 July 2007), pp. 7619-7630.

Electrical microstimulation is used widely in experimental neurophysiology to examine causal links between specific brain areas and their behavioral functions and is used clinically to treat neurological and psychiatric disorders in patients. Typically, microstimulation is applied to local brain regions as a train of equally spaced current pulses. We were interested in the sensitivity of a neural circuit to a train of variably spaced pulses, as is observed in physiological spike trains. We compared the effect of fixed, decelerating, accelerating, and randomly varying microstimulation patterns on the likelihood and metrics of eye movements evoked from the frontal eye field of monkeys, while holding the mean interpulse interval constant. Our results demonstrate that the pattern of microstimulation pulses strongly influences the probability of evoking a saccade, as well as the metrics of the saccades themselves. Specifically, the pattern most closely resembling physiological spike trains (accelerating pattern) was most effective at evoking a saccade, three times more so than the least effective decelerating pattern. A saccade-triggered average of effective random trains confirmed the positive relationship between accelerating rate and efficacy. These results have important implications for the use of electrical microstimulation in both experimental and clinical settings and suggest a means to study the role of temporal pattern in the encoding of behavioral and cognitive functions. 10.1523/JNEUROSCI.0386-07.2007