PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • All procedures were approved by the University Committee on Animal Resources at the University of Rochester and were designed and conducted in compliance with the Public Health Service’s Guide for the Care and Use of Animals (protocol UCAR-2010-169). Two male rhesus macaques (Macaca mulatta, subject B: age, 8 years, 11 months; subject J: age, 10 years, 9 months) served as subjects. A small prosthesis for holding the head was used. Animals were habituated to laboratory conditions and then trained to perform oculomotor tasks for liquid reward. A Cilux recording chamber (Crist Instruments) was placed over the dACC. Position was verified by magnetic resonance imaging with the aid of a Brainsight system (Rogue Research Inc.). Animals received appropriate analgesics and antibiotics after all procedures. Throughout both behavioral and physiological recording sessions, the chamber was kept sterile with regular antibiotic washes and sealed with sterile caps. All recordings were performed during the animals’ light cycle, between 8 AM and 5 PM. Some of the data for dACC recordings were previously published [42]; all analyses presented here are new. We approached the dACC through a standard recording grid (Crist Instruments). We defined the dACC according to the Paxinos atlas [72]. Roughly, we recorded from a region of interest lying within the coronal planes situated between 29.50 and 34.50 mm rostral to interaural plane, the horizontal planes situated between 4.12 and 7.52 mm from the brain’s dorsal surface, and the sagittal planes situated between 0 and 5.24 mm from medial wall. The atlas called these Areas 8/32 and 9/32; we prefer to call them Area 24 [38]. Our recordings were made from a central region within this zone. We confirmed recording location before each recording session using our Brainsight system with structural magnetic resonance images taken before the experiment. Neuroimaging was performed at the Rochester Center for Brain Imaging on a Siemens 3T MAGNETOM Trio Tim using 0.5-mm voxels. We confirmed recording locations by listening for characteristic sounds of white and gray matter during recording, which in all cases matched the loci indicated by the Brainsight system. Electrophysiological techniques, eye tracking, and reward delivery: All methods used were described in previous manuscripts [40] and largely reproduced here. Single electrodes (Frederick Haer & Co., impedance range 0.8–4 MU) were lowered using a microdrive (NAN Instruments) until waveforms of between 1 and 3 neuron(s) were isolated. Individual action potentials were isolated on a Plexon system (Plexon, Inc.). Neurons were selected for study solely on the basis of the quality of isolation; we never preselected based on task-related response properties. All collected neurons for which we managed to obtain at least 250 trials were analyzed. Eye position was sampled at 1,000 Hz by an infrared eye-monitoring camera system (SR Research). Stimuli were controlled by a computer running Matlab (Mathworks) with Psychtoolbox [73] and Eyelink Toolbox [74]. Visual stimuli were colored rectangles on a computer monitor placed 57 cm from the animal and centered on its eyes (Fig 2A). A standard solenoid valve controlled the duration of juice delivery. The relationship between solenoid open time and juice volume was established and confirmed before, during, and after recording. Monkeys performed a 2-option gambling task. The task was similar to one we have used previously [40,41,75], with 2 major differences: first, monkeys gambled for virtual tokens rather than liquid rewards, and second, outcomes could be losses as well as wins. Two offers were presented on each trial. Each offer was represented by a rectangle 300 pixels tall and 80 pixels wide (11.35° of the visual angle tall and 4.08° of the visual angle wide). Twenty percent of options were safe (100% probability of either 0 or 1 token), while the remaining 80% were gambles. Safe offers were entirely red (0 tokens) or blue (1 token). The size of each portion indicated the probability of the respective reward. Each gamble rectangle was divided horizontally into a top and bottom portion, each colored according to the token reward offered. Gamble offers were thus defined by 3 parameters: 2 possible token outcomes, and the probability of the top outcome (the probability of the bottom was strictly determined by the probability of the top). The top outcome was 10%, 30%, 50%, 70%, or 90% likely. Six initially unfilled circles arranged horizontally at the bottom of the screen indicated the number of tokens to be collected before the subject obtained a liquid reward. These circles were filled appropriately at the end of each trial, according to the outcome of that trial. When 6 or more tokens were collected, the tokens were covered with a solid rectangle while a liquid reward was delivered. Tokens beyond 6 did not carry over nor could number of tokens fall below zero. On each trial, one offer appeared on the left side of the screen and the other appeared on the right. Offers were separated from the fixation point by 550 pixels (27.53° of the visual angle). The side of the first offer (left and right) was randomized by trial. Each offer appeared for 600 ms and was followed by a 150-ms blank period. Monkeys were free to fixate upon the offers when they appeared (and in our observations almost always did so). After the offers were presented separately, a central fixation spot appeared and the monkey fixated on it for 100 ms. Following this, both offers appeared simultaneously and the animal indicated its choice by shifting gaze to its preferred offer and maintaining fixation on it for 200 ms. Failure to maintain gaze for 200 ms did not lead to the end of the trial but instead returned the monkey to a choice state; thus, monkeys were free to change their mind if they did so within 200 ms (although in our observations, they seldom did so). A successful 200-ms fixation was followed by a 750-ms delay, after which, the gamble was resolved and a small reward (100 μL) was delivered—regardless of the outcome of the gamble—to sustain motivation. This small reward was delivered within a 300-ms window. If 6 tokens were collected, a delay of 500 ms was followed by a large liquid reward (300 μL) within a 300-ms window, followed by a random intertrial interval (ITI) between 0.5 and 1.5 s. If 6 tokens were not collected, subjects proceeded immediately to the ITI. Each gamble included at least one positive or zero outcome. This decreased the number of trivial choices presented to subjects and maintained motivation. Subjective values for each gamble were estimated based on subjects’ behavior performance in each test session, according to the following formula: SV=p*winα+(1−p)*lossβ (the same approach used by Yamada and colleagues [76]). Because our task includes both wins and losses, we fit a parameter α for wins and another parameter β for losses. A value for α greater than 1 and a value for β less than 1 both indicate risk seeking. Both subjects were risk seeking on average (values of α > 1 or β < 1 both indicate risk seeking; subject B: average α = 1.21, SD = 0.409, average β = 0.0764, SD = 0.132; subject J: average α = 1.60, SD = 0.404, average β = 0.0216, SD = 0.0530). We also fit subjective values conditioned on the number of tokens the subject had accumulated as of the beginning of each trial. We did this by fitting the above equation to trials in each possible tokens-accumulated condition (accumulated tokens = 0, 1, …, 5). We thus obtained 6 different parameters for gains and losses for each subject (S2 Table). We used the subjective values fit by the parameters obtained using these 2 methods to replicate our analyses of significantly-modulated neurons (S3 Table) and beta correlation analyses (S4 Table). We fit logistic regression models of behavior to predict choice of the first versus second offer. To ensure that subjects did, in fact, pay attention to both offers, we fit a model in which the values of the first and second offers were the predictors of interest, while also including the number of tokens already accumulated, the side the first offer appears on, and the choice eventually made to explain any variance these variables might contribute to. To determine whether subjects pay attention to all features of an offer, we use an extended model with the 3 variables characterizing each offer (the 2 possible outcomes and the probability of the larger outcome) included as predictors, controlling for the same variables mentioned above. We fit such a model for each behavioral session and obtained the regression weights associated with each of the variables of interest. We then tested the vector of these variables across all sessions using a 1-sample t test to determine whether they differ significantly from zero. Peristimulus time histograms (PSTHs) were constructed by aligning spike rasters to the presentation of the first offer and averaging firing rates across multiple trials. Firing rates were calculated in 20-ms bins but were generally analyzed in longer (500-ms) epochs. This method is standard in our lab and was described in a previous manuscript [40]. Firing rates were normalized by subtracting the mean and dividing by the standard deviation of the entire neuron’s PSTH. We tested for significant neuronal modulation using a multiple linear regression, including the following task-relevant variables: expected/subjective value of offers 1 and 2, the number of tokens collected as of the beginning of the trial, the side the first offer appeared on, and the side of the chosen offer. Analysis epochs were chosen a priori before data analysis began to reduce the likelihood of p-hacking. The first and second offer epochs were defined as the 500-ms epoch beginning 100 ms after the offer was presented, to account for information-processing time. These epochs were used in previous studies of choice behavior [40,41]. All fractions of modulated neurons were tested for significance using a 2-sided binomial test. All binomial tests throughout the manuscript were 2-sided. Format and population correlation analyses: We used beta correlation analyses to assess whether neurons represented 2 variables (or the same variable at different time periods) using similar/orthogonal/opposing formats, in overlapping/orthogonal/distinct populations. To do this, we first found the regression coefficient associated with the variable of interest (controlling for other task variables, listed above) per neuron. We then combined these regression coefficients into a vector of the same length as the number of neurons in our sample. This vector indicates the strength and direction of modulation for each individual neuron in the population, in response to a particular variable in a particular epoch. We call this the population “format.” We compared different formats by finding the Pearson correlation coefficient between them. We modify this method slightly to account for the noise levels of each individual neuron’s encoding of each variable of interest, which our existing method cannot account for. We used a Bayesian regression to obtain a probabilistic distribution over each regression coefficient for each neuron, rather than an individual value per neuron. We sampled 10,000 regression coefficients from this distribution per neuron to obtain 10,000 potential formats for the population. We then performed the correlation analyses on each of these samples, thus generating a distribution of 10,000 correlation coefficients. This is a more robust estimate of the correlation between formats, as it takes into account the uncertainty inherent in estimating any individual regression coefficient and allows us to view the spread of the distribution of this correlation when this significant source of noise is taken into account. Credible intervals of 99% allowed us to estimate the likely range of the correlation coefficients with 99% certainty. The Pearson correlation coefficient between signed regression coefficients indicates whether variables were represented in a similar format, i.e., directionality of tuning across the population. A positive correlation indicates a preservation of directionality, while a negative correlation suggests variables were represented in opposing directionality of firing rate modulation. No correlation suggests orthogonal formats. Similarly, the Pearson correlation coefficient between unsigned regression coefficients indicates whether similar neuronal populations tended to be involved in encoding the 2 variables in question. A positive correlation indicates overlapping populations, while a negative correlation indicates separate ones. A lack of correlation suggests orthogonal populations (i.e., encoding 1 variable does not affect the neuron’s likelihood of encoding the other variable). We then compare these distributions of correlation coefficients to distributions that would be obtained under a chance model. For the first set of analyses, meant to differentiate between the 1-pool and 2-pool hypotheses, we assume a chance model is one in which, during the second epoch, neurons encode the values of both offers but don’t differentiate between them. We achieve this by shuffling the values of the 2 offers and using these new shuffled vectors as predictors in our regression model. For the second set of analyses, examining the encodings of accepted and rejected offers, we assume a chance model in which neurons do not differentiate between values according to whether they will later be accepted or rejected. To achieve this, we shuffle trials across these 2 categories (offer 1 or 2 accepted/rejected) at random. These chance models achieve a permutation of the existing data, which we then use for the same beta correlation analyses explained above. We then compare the mean correlation of our actual data to the distribution of mean correlations obtained from each of 1,000 permutations of our data. This form of test allows us to (1) ensure that we have enough signal in our data to detect significant correlations, and (2) determine whether the variable of interest does, in fact, play a role in differentiating the formats/populations involved. As we have often found in this study, a positive correlation between formats, while in itself informative, may not actually be as strong as would be expected purely by chance given this dataset, which is also, in itself, an important finding that informs our theories and interpretations. Data were deposited in the Dryad depository http://dx.doi.org/10.5061/dryad.h52f8 [77].
rdf:type