FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2019909176> ?p ?o ?g. }

• W2019909176 endingPage "1416" @default.
• W2019909176 startingPage "1401" @default.
• W2019909176 abstract "•PMd and M1 are involved in deliberation between action choices during dynamic tasks•PMd and M1 track the state of information and combine it with an urgency signal•PMd and M1 activity signals decision commitment about 280 ms before movement•The neural signature of commitment is specific to volitional decisions Neurophysiological studies of decision making have primarily focused on decisions about information that is stable over time. However, during natural behavior, animals make decisions in a constantly changing environment. To investigate the neural mechanisms of such dynamic choices, we recorded activity in dorsal premotor (PMd) and primary motor cortex (M1) while monkeys performed a two-choice reaching task in which sensory information about the correct choice was changing within each trial and the decision could be made at any time. During deliberation, activity in both areas did not integrate sensory information but instead tracked it and combined it with a growing urgency signal. Approximately 280 ms before movement onset, PMd activity tuned to the selected target reached a consistent peak while M1 activity tuned to the unselected target was suppressed. We propose that this reflects the resolution of a competition between the potential responses and constitutes the volitional commitment to an action choice. Neurophysiological studies of decision making have primarily focused on decisions about information that is stable over time. However, during natural behavior, animals make decisions in a constantly changing environment. To investigate the neural mechanisms of such dynamic choices, we recorded activity in dorsal premotor (PMd) and primary motor cortex (M1) while monkeys performed a two-choice reaching task in which sensory information about the correct choice was changing within each trial and the decision could be made at any time. During deliberation, activity in both areas did not integrate sensory information but instead tracked it and combined it with a growing urgency signal. Approximately 280 ms before movement onset, PMd activity tuned to the selected target reached a consistent peak while M1 activity tuned to the unselected target was suppressed. We propose that this reflects the resolution of a competition between the potential responses and constitutes the volitional commitment to an action choice. When buying a house, one is motivated to first collect relevant information and then take time to think about the best choice. Because careful deliberation is important to human behavior, studies of the neural mechanisms of decision making have largely focused on scenarios in which subjects decide about information that is stable over time. For example, perceptual decisions are usually studied using stimuli whose informational content is constant in each trial (Britten et al., 1992Britten K.H. Shadlen M.N. Newsome W.T. Movshon J.A. The analysis of visual motion: a comparison of neuronal and psychophysical performance.J. Neurosci. 1992; 12: 4745-4765Crossref PubMed Google Scholar, Romo et al., 2004Romo R. Hernández A. Zainos A. Neuronal correlates of a perceptual decision in ventral premotor cortex.Neuron. 2004; 41: 165-173Abstract Full Text Full Text PDF PubMed Scopus (265) Google Scholar), leading to models of deliberation as the integration of sensory evidence to a threshold (Gold and Shadlen, 2007Gold J.I. Shadlen M.N. The neural basis of decision making.Annu. Rev. Neurosci. 2007; 30: 535-574Crossref PubMed Scopus (2264) Google Scholar, Ratcliff, 1978Ratcliff R. A theory of memory retrieval.Psychol. Rev. 1978; 85: 59-108Crossref Scopus (2982) Google Scholar). Likewise, studies of value-based decisions focus on conditions in which the value of options is stable (Padoa-Schioppa, 2011Padoa-Schioppa C. Neurobiology of economic choice: a good-based model.Annu. Rev. Neurosci. 2011; 34: 333-359Crossref PubMed Scopus (384) Google Scholar, Platt and Glimcher, 1999Platt M.L. Glimcher P.W. Neural correlates of decision variables in parietal cortex.Nature. 1999; 400: 233-238Crossref PubMed Scopus (1218) Google Scholar), leading to serial models in which the costs and benefits are converted into a “common currency,” the decision is made, and the chosen action is then prepared (Padoa-Schioppa, 2011Padoa-Schioppa C. Neurobiology of economic choice: a good-based model.Annu. Rev. Neurosci. 2011; 34: 333-359Crossref PubMed Scopus (384) Google Scholar). However, the vertebrate brain evolved to guide behavior in a dynamic world, in which decisions are made during ongoing activity, action options and their payoffs are continuously changing, and animals are free to decide when to take time to deliberate and when to commit quickly to their current best guess. Here, we investigate such “embodied” decisions and ask which conclusions from static scenarios generalize to real-time dynamic decisions. In particular, studies of static tasks have suggested that the brain gradually integrates repeated samples of the stimulus, causing neural activity to build up to a threshold (Gold and Shadlen, 2007Gold J.I. Shadlen M.N. The neural basis of decision making.Annu. Rev. Neurosci. 2007; 30: 535-574Crossref PubMed Scopus (2264) Google Scholar, Ratcliff, 1978Ratcliff R. A theory of memory retrieval.Psychol. Rev. 1978; 85: 59-108Crossref Scopus (2982) Google Scholar). However, if the sensory information can suddenly change, such a process is suboptimal, because integrators are sluggish to respond to changes in input. Recent human studies of dynamic tasks have suggested that instead of integrating the sensory state, the brain quickly tracks it, and activity buildup is caused by a growing urgency to act (Cisek et al., 2009Cisek P. Puskas G.A. El-Murr S. Decisions in changing conditions: the urgency-gating model.J. Neurosci. 2009; 29: 11560-11571Crossref PubMed Scopus (270) Google Scholar, Thura et al., 2012Thura D. Beauregard-Racine J. Fradet C.W. Cisek P. Decision making by urgency gating: theory and experimental support.J. Neurophysiol. 2012; 108: 2912-2930Crossref PubMed Scopus (155) Google Scholar). These models can only be distinguished with dynamic tasks, because they make identical predictions for any static task, at both the behavioral and neural level. Furthermore, many neurophysiological studies have shown that decision making influences activity in the sensorimotor system (Gold and Shadlen, 2000Gold J.I. Shadlen M.N. Representation of a perceptual decision in developing oculomotor commands.Nature. 2000; 404: 390-394Crossref PubMed Scopus (441) Google Scholar, Platt and Glimcher, 1999Platt M.L. Glimcher P.W. Neural correlates of decision variables in parietal cortex.Nature. 1999; 400: 233-238Crossref PubMed Scopus (1218) Google Scholar, Salinas and Romo, 1998Salinas E. Romo R. Conversion of sensory signals into motor commands in primary motor cortex.J. Neurosci. 1998; 18: 499-511Crossref PubMed Google Scholar, Wallis and Miller, 2003Wallis J.D. Miller E.K. From rule to response: neuronal processes in the premotor and prefrontal cortex.J. Neurophysiol. 2003; 90: 1790-1806Crossref PubMed Scopus (268) Google Scholar). In particular, when animals are faced with multiple response options, the brain represents them in parallel within sensorimotor regions (Baumann et al., 2009Baumann M.A. Fluet M.C. Scherberger H. Context-specific grasp movement representation in the macaque anterior intraparietal area.J. Neurosci. 2009; 29: 6436-6448Crossref PubMed Scopus (184) Google Scholar, Cisek and Kalaska, 2005Cisek P. Kalaska J.F. Neural correlates of reaching decisions in dorsal premotor cortex: specification of multiple direction choices and final selection of action.Neuron. 2005; 45: 801-814Abstract Full Text Full Text PDF PubMed Scopus (701) Google Scholar, McPeek et al., 2003McPeek R.M. Han J.H. Keller E.L. Competition between saccade goals in the superior colliculus produces saccade curvature.J. Neurophysiol. 2003; 89: 2577-2590Crossref PubMed Scopus (186) Google Scholar), and these representations are modulated by decision variables (Basso and Wurtz, 1998Basso M.A. Wurtz R.H. Modulation of neuronal activity in superior colliculus by changes in target probability.J. Neurosci. 1998; 18: 7519-7534Crossref PubMed Google Scholar, Dorris and Glimcher, 2004Dorris M.C. Glimcher P.W. Activity in posterior parietal cortex is correlated with the relative subjective desirability of action.Neuron. 2004; 44: 365-378Abstract Full Text Full Text PDF PubMed Scopus (340) Google Scholar, Pastor-Bernier and Cisek, 2011Pastor-Bernier A. Cisek P. Neural correlates of biased competition in premotor cortex.J. Neurosci. 2011; 31: 7083-7088Crossref PubMed Scopus (118) Google Scholar, Roitman and Shadlen, 2002Roitman J.D. Shadlen M.N. Response of neurons in the lateral intraparietal area during a combined visual discrimination reaction time task.J. Neurosci. 2002; 22: 9475-9489Crossref PubMed Google Scholar, Yang and Shadlen, 2007Yang T. Shadlen M.N. Probabilistic reasoning by neurons.Nature. 2007; 447: 1075-1080Crossref PubMed Scopus (368) Google Scholar). For example, information for deciding between manual actions influences neural activity in premotor and parietal regions (Hernández et al., 2010Hernández A. Nácher V. Luna R. Zainos A. Lemus L. Alvarez M. Vázquez Y. Camarillo L. Romo R. Decoding a perceptual decision process across cortex.Neuron. 2010; 66: 300-314Abstract Full Text Full Text PDF PubMed Scopus (206) Google Scholar, Klaes et al., 2011Klaes C. Westendorff S. Chakrabarti S. Gail A. Choosing goals, not rules: deciding among rule-based action plans.Neuron. 2011; 70: 536-548Abstract Full Text Full Text PDF PubMed Scopus (108) Google Scholar, Pastor-Bernier and Cisek, 2011Pastor-Bernier A. Cisek P. Neural correlates of biased competition in premotor cortex.J. Neurosci. 2011; 31: 7083-7088Crossref PubMed Scopus (118) Google Scholar), modulates corticospinal excitability (Klein-Flügge and Bestmann, 2012Klein-Flügge M.C. Bestmann S. Time-dependent changes in human corticospinal excitability reveal value-based competition for action during decision processing.J. Neurosci. 2012; 32: 8373-8382Crossref PubMed Scopus (79) Google Scholar, Michelet et al., 2010Michelet T. Duncan G.H. Cisek P. Response competition in the primary motor cortex: corticospinal excitability reflects response replacement during simple decisions.J. Neurophysiol. 2010; 104: 119-127Crossref PubMed Scopus (60) Google Scholar), and even influences reflexes (Selen et al., 2012Selen L.P. Shadlen M.N. Wolpert D.M. Deliberation in the motor system: reflex gains track evolving evidence leading to a decision.J. Neurosci. 2012; 32: 2276-2286Crossref PubMed Scopus (133) Google Scholar). Such results have led to the proposal that decisions between actions involve processes within the sensorimotor system (Cisek, 2007Cisek P. Cortical mechanisms of action selection: the affordance competition hypothesis.Philos. Trans. R. Soc. Lond. B Biol. Sci. 2007; 362: 1585-1599Crossref PubMed Scopus (658) Google Scholar, Hernández et al., 2010Hernández A. Nácher V. Luna R. Zainos A. Lemus L. Alvarez M. Vázquez Y. Camarillo L. Romo R. Decoding a perceptual decision process across cortex.Neuron. 2010; 66: 300-314Abstract Full Text Full Text PDF PubMed Scopus (206) Google Scholar, Shadlen et al., 2008Shadlen M.N. Kiani R. Hanks T.D. Churchland A.K. Neurobiology of Decision Making: An Intentional Framework.in: Engel C. Singer W. Better Than Conscious? Decision Making, the Human Mind, and Implications for Institutions. MIT press, Cambridge2008: 71-101Crossref Google Scholar). However, it is also possible that they simply reflect information that spills in from upstream regions that are actually responsible for deliberation and commitment. To establish whether the sensorimotor system plays an active role in deliberation, neural activity must be examined well before commitment is made, and this can be accomplished with dynamic tasks. Here, we investigate these questions through neurophysiological recordings in the dorsal premotor (PMd) and primary motor cortex (M1) of monkeys trained to perform two reaching tasks. In the “tokens” task (Figure 1A), monkeys watch a set of 15 tokens jumping every 200 ms from a central target to one of two peripheral targets and must guess which target will ultimately receive the majority of the tokens. Importantly, the decision can be taken at any time, and when a target is reached, the token jumps accelerate, allowing the monkey to save time by taking an early guess. In the “delayed response” (DR) task, only a single peripheral target is presented, and the monkey must withhold movement until the 15 tokens jump into the target simultaneously (GO signal). Our paradigm has two critical properties. First, the sensory evidence in the tokens task is continuously changing, allowing us to dissociate different models of how sensory information is treated. Second, in the tokens task, the monkeys are free to respond at any time, allowing us to distinguish processes related to deliberation from those related to commitment. In particular, by comparing activity during the tokens task with activity during the DR task, in which both the choice and its timing are externally instructed, we can identify the neural phenomena specifically associated with volitional commitment to action. In the tokens task, monkeys’ success rate varied between 64%–87% (mean: 77%; SD: 5%, SE: 0.5%). To analyze how behavior depended upon the specific pattern of token jumps in each trial, we first estimated the total sensory and motor delays using the mean reaction time (mRT) from the DR task (Monkey S: 291 ± 40 ms; Monkey Z: 335 ± 93 ms) and then subtracted this from the reaction time (RT) in the tokens task to estimate the decision time (DT) (Figure 1B). Next, we estimated the success probability at decision time (SPD) using Equation 1 below. We compared these variables in three trial types: easy, ambiguous, and misleading (Figure 1C), classified post hoc from the fully random trials. As expected, both monkeys made decisions significantly earlier in easy than in ambiguous or misleading trials (Kolmogorov-Smirnov [KS] test, p < 0.01) (Figure 1D). They also made decisions at a significantly lower level of success probability in ambiguous and misleading trials than in easy trials (KS test, p < 0.05) (Figure 1E). This is consistent with the hypothesis that to solve the task, monkeys use an accuracy criterion that decreases over time. To test this across all trials, we grouped data according to the number of tokens that moved before DT and calculated an estimate of the accuracy criterion (or “confidence”) for the selected target at that time. This estimate was based on the sum of the log likelihood ratios of individual token jumps (SumLogLR; see Experimental Procedures), which is related to the difference in the number of tokens in the two targets. The result is shown in Figure 1F. Except for fast guesses (<1 s), there is a trend for decisions to be made at a lower level of accuracy as time passes. This demonstrates that both monkeys use a similar strategy as humans to solve the task (Cisek et al., 2009Cisek P. Puskas G.A. El-Murr S. Decisions in changing conditions: the urgency-gating model.J. Neurosci. 2009; 29: 11560-11571Crossref PubMed Scopus (270) Google Scholar)—they decrease their accuracy criterion over time. Previous studies have suggested that this can be implemented by combining sensory information with a growing “urgency signal” (Churchland et al., 2008Churchland A.K. Kiani R. Shadlen M.N. Decision-making with multiple alternatives.Nat. Neurosci. 2008; 11: 693-702Crossref PubMed Scopus (437) Google Scholar, Cisek et al., 2009Cisek P. Puskas G.A. El-Murr S. Decisions in changing conditions: the urgency-gating model.J. Neurosci. 2009; 29: 11560-11571Crossref PubMed Scopus (270) Google Scholar, Standage et al., 2011Standage D. You H. Wang D.H. Dorris M.C. Gain modulation by an urgency signal controls the speed-accuracy trade-off in a network model of a cortical decision circuit.Front. Comput. Neurosci. 2011; 5: 7Crossref PubMed Scopus (49) Google Scholar, Thura et al., 2012Thura D. Beauregard-Racine J. Fradet C.W. Cisek P. Decision making by urgency gating: theory and experimental support.J. Neurophysiol. 2012; 108: 2912-2930Crossref PubMed Scopus (155) Google Scholar). While monkeys performed the tokens task, neural activity was recorded from 178 cells in the arm area of PMd (135 in monkey S) and 74 cells in M1 (55 in monkey S) excluding the most caudal region in the central sulcus (Figure S1 available online). Among these, 99 cells (68 in PMd and 31 in M1) had a significant directional preference before DT (see Experimental Procedures) and thus reliably predicted whether an arm movement would be made toward or away from the cell’s preferred target (PT). Figure 2A shows the activity of an example PMd neuron, aligned on the start of token jumps and plotted until 300 ms before movement onset, during easy, ambiguous, and misleading trials in which the monkey chose the cell’s PT or the opposite target (OT). In easy trials, activity strongly and quickly increases when the monkey selects the PT and is quickly suppressed when the OT is chosen. In ambiguous trials, activity fluctuates and gradually increases during the first seven token jumps (∼1.4 s), and the cell discriminates between PT and OT only late in the trial. The pattern of activity is again very different in misleading trials. When the PT is ultimately selected, activity is initially low while the first few tokens favor the OT and later switches to predict the PT choice. In contrast, in OT trials the early activity is strong, while tokens favor the PT, and later decreases. Interestingly, we found very similar phenomena in M1, as shown in Figure 2B for an example cell. Like the PMd cell in Figure 2A, this M1 neuron quickly reflected the choice in easy trials, fluctuated during ambiguous trials, and reflected the switch of evidence in misleading trials. Figure 2C illustrates the average activity of all 68 PMd and 31 M1 decision-related neurons. The top panel shows the profile of success probability for each cell’s PT for easy (blue), ambiguous (green), and misleading (red) trials in which the monkey correctly chose the PT (solid) or OT (dashed). Below that, we show the average neural activity of the 68 PMd and 31 M1 cells aligned on the first token jump and plotted until 300 ms before movement initiation (diamonds) during those same trials. About 150 ms after the first token jump, activity increases or decreases in a manner that reflects the sensory evidence and the monkey’s ultimate choice, especially in PMd. In addition to the influence of the changing sensory evidence, there is a trend for activity to increase over time, and this is especially pronounced in M1. To further quantify these observations, we calculated the latency at which activity discriminates between PT and OT for all PMd and M1 decision-related cells in easy or ambiguous trials. The mean discrimination time was significantly shorter in easy than in ambiguous trials both in PMd (280 versus 624 ms; KS test; p < 0.01) and in M1 (341 versus 807 ms; KS test; p < 0.01). Moreover, in both easy and ambiguous trials, discrimination times were shorter in PMd than in M1, although this did not reach significance (KS test; p > 0.05). To quantify the effect of sensory evidence on neural activity across all trials (not just the three types shown in Figure 2), we measured each cell’s firing rate in successive 200 ms epochs following the first token jump. We then plotted this as a function of the sensory evidence (SumLogLR) present during the previous token jump (to allow for sensory delays). The result for example PMd and M1 cells is shown in Figures 3A and 3C, respectively. Both neurons exhibited a clear modulation of activity as a function of the evidence, increasing their firing rate as the evidence in favor of their PT increased (Spearman’s rank test; mean r = 0.93 for the PMd neuron and 0.75 for the M1 neuron). Moreover, this relationship changed as time was passing. In particular, both the baseline and the slope (calculated at the equal evidence point; vertical dashed line) tended to grow over time. These effects also held at the population level in both PMd and M1 (Figures 3B and 3D). Notably, at the population level, the increase of activity over time was primarily due to a baseline shift (Figure S2). The above results raise the question of what computational mechanism transforms sensory information into neural activity and what is responsible for the activity buildup. It has been suggested that during perceptual discrimination, such as deciding about the direction of noisy motion, the sensory signal is integrated over time by repeatedly resampling the stimulus (Bogacz et al., 2006Bogacz R. Brown E. Moehlis J. Holmes P. Cohen J.D. The physics of optimal decision making: a formal analysis of models of performance in two-alternative forced-choice tasks.Psychol. Rev. 2006; 113: 700-765Crossref PubMed Scopus (1146) Google Scholar, Gold and Shadlen, 2007Gold J.I. Shadlen M.N. The neural basis of decision making.Annu. Rev. Neurosci. 2007; 30: 535-574Crossref PubMed Scopus (2264) Google Scholar). Does a similar mechanism operate during the tokens task? In particular, does the brain temporally integrate the state of the sensory information about which the decision is made (i.e., the distribution of tokens), or does it simply track that sensory state? To distinguish between these two mechanisms, we examined additional trial types, classified post hoc from the fully random set. For example, during “bias-up/down” (BUD) trials (Figure 4A, top, green line) the first three tokens move to the PT, then the next two move toward the OT, and then the rest of the trial resembles an easy trial toward the PT (which is the correct target). In contrast, during “bias-down/up” (BDU) trials, the first two tokens move to the OT and the next three to the PT, and the rest of the trial is similar to BUD. Comparison between these two trial types is critical, because if the sensory state is integrated, then after the fifth token neural activity related to the PT will be higher in BUD than in BDU trials (because an integrator retains a “memory” of previous states). In contrast, if the sensory state is simply tracked, then one would not predict a significant difference between the trial types. Figure 4A illustrates the activity of one PMd neuron recorded during BUD and BDU trials. Importantly, we include only trials in which the monkey made decisions after the initial bias (>800 ms) and truncate activity 300 ms before movement onset. About 200 ms after the first token jump, activity begins to reflect the bias in the two trial types, becoming stronger in BUD than BDU trials. After 800 ms, as sensory evidence converges in both trials, neural activity likewise becomes similar. Therefore, this neuron did not integrate the sensory state during the bias, but instead tracked it quickly (note the rapid increase of activity after 800 ms in BDU trials, clearly visible in the rasters). This observation holds true when activity is averaged across PMd and M1 cells (Figure 4B). To test whether this effect is robust across individual cells, we compared the mean activity of each PMd and M1 decision-related cell in BUD and BDU trials in two epochs: during and after the bias. As predicted by both mechanisms, during the bias, the response is usually stronger in BUD than BDU trials. However, after the bias, most of the cells no longer discharge differently in the two trial types, consistent with a system that simply tracks the sensory state (Figures 4C and 4D). The integration and tracking mechanisms also make very distinct predictions at a behavioral level. In particular, integration predicts faster decisions in BUD than BDU trials, whereas tracking predicts no difference. In agreement with the latter, we found no statistical difference between decision durations in BUD versus BDU trials in both monkeys (Figure 5A, top row). Similar analyses on two other pairs of trial types (Figure 5A, middle and bottom rows) yielded the same conclusion—that choices were not biased by the early evidence, consistent with fast tracking of the sensory state and not with integration. In Figure S3 we show that neural activity in both PMd and M1 also tracks the sensory state in these trial types and exhibits no memory of the sensory state after the initial bias has ended. The observations above suggest that the sensory information is quickly tracked, perhaps through a low-pass filter with a very short time window (i.e., less than the duration between two token jumps [200 ms]). To test this explicitly, we computed the mean neural activity during two 200 ms epochs, from 200–400 ms and from 400–600 ms after the first token jump, and sorted trials according to the pattern of the two first token jumps: in step-for trials, the first token jumps to the cells’ PT, and the second token jumps to the OT. In step-against trials, this pattern is reversed. Accounting for sensory delays, the first epoch of analysis reflects the consequences of the first token jump. Comparison between step-for and step-against trials shows that activity in PMd is significantly higher (KS test; p < 0.05) and shows a similar trend in M1 during that first epoch in step-for than in step-against trials (Figure 5B). However, 200 ms later in the trial, activity is similar between the two conditions, in agreement with fast tracking of sensory information. Because monkeys are allowed to make their decision at any time during the tokens task, we can examine cortical activity for the neural signature of the moment at which commitment to a choice is made. To this end, we aligned activity on movement onset, as shown on Figures 6A and 6B, for one PMd and one M1 neuron. Note that approximately 300 ms before the monkey chooses the cell’s PT, there is a clear peak of activity in both cells, and the amplitude and timing of this peak are similar across the different trial types. As shown in Figure S4, this phenomenon exists at the level of individual trials and is not an artifact of averaging over trials of different lengths. It is also not related to saccades: although analysis of oculomotor behavior in our task will be described in a future publication, it is important to mention here that in most trials (74%–79%), the monkeys were already fixating the chosen target well before the peak in PMd activity. Figure 6C shows the average activity of the 68 PMd and 31 M1 neurons aligned on movement onset. In both areas, neural activity related to the PT shows a striking characteristic regardless of the trial type: between the start of token jumps and movement onset, activity first shows the influence of the mounting sensory evidence, then reaches a peak, and finally decreases prior to movement onset. The approximate timing of that peak (vertical gray line in Figure 6C) as well as its amplitude is very similar across trial types, appearing earlier in PMd than in M1. Importantly, at approximately the same time, activity in M1 related to the unselected OT target becomes rapidly suppressed. This is seen most clearly in ambiguous and misleading trials, in which there was some evidence favoring the OT choice. To further quantify the timing of these phenomena, we first averaged the activity of PMd and M1 neurons across all trials in which the PT was chosen and detected the peak firing rate across 10 ms bins. In PMd, this peak was reached 280 ms before movement onset, whereas in M1 it occurred 140 ms later (Figures 7A and 7D, left). To assess the robustness of this observation, we also calculated the peak timing for each cell separately and computed its mean and median latency across the population (Figures 7A and 7D, right). Except for a few cells whose maximum activity occurs very early, there is a clear trend for the peak to occur approximately 260 ms before movement onset in PMd and 179 ms in M1. Next, we sought to determine whether the latencies of neural activity peaks correlate with RTs and are consistent across trial types. For each cell, we calculated the mean latency of the peak in easy and misleading trials and then plotted these against the RT in those same trials (Figures 7B and 7E). In almost all PMd and M1 neurons, there is a strong relationship with a slope near unity, suggesting that regardless of the history of sensory information during a trial, the timing of PMd and M1 peaks is consistent relative to movement initiation. Here, we will use the earliest of these latency values as our estimate of the timing of commitment: 280 ms before movement onset. Figure 7C shows that the amplitude of the PMd activity at this moment in PT trials is very consistent across trial types, although there is a slight trend for it to be lower in easy trials, in which decisions are shorter. Most interestingly, at this same moment, M1 activity tuned to the unselected target becomes rapidly suppressed (Figure 6C). For each PMd and M1 cell, we quantified the timing of this suppression by examining how activity changed between two consecutive 50 ms bins at different latencies with respect to movement onset (see Figure S5). The distribution of the time of the greatest change between bins is quite broad for PMd, suggesting that there is no specific moment of suppression of cells tuned to the unselected target. In M1, however, the distribution is narrower, and the mean timing across cells (275 ms) roughly corresponds to our estimate of commitment time. In the tokens task, monkeys are free to decide both which target to choose and the time at which they commit to that choice. Figures 6 and 7 suggest that the neural peak in PMd and suppression in M1 signals this moment of free commitment. However, are these phenomena simply th" @default.
• W2019909176 created "2016-06-24" @default.
• W2019909176 creator A5018287398 @default.
• W2019909176 creator A5089018845 @default.
• W2019909176 date "2014-03-01" @default.
• W2019909176 modified "2023-10-06" @default.
• W2019909176 title "Deliberation and Commitment in the Premotor and Primary Motor Cortex during Dynamic Decision Making" @default.
• W2019909176 cites W1487392737 @default.
• W2019909176 cites W1579888686 @default.
• W2019909176 cites W1589439850 @default.
• W2019909176 cites W1609623940 @default.
• W2019909176 cites W1821259251 @default.
• W2019909176 cites W1975842858 @default.
• W2019909176 cites W1979224386 @default.
• W2019909176 cites W1981330394 @default.
• W2019909176 cites W1990260016 @default.
• W2019909176 cites W1993594397 @default.
• W2019909176 cites W1998171734 @default.
• W2019909176 cites W2006767346 @default.
• W2019909176 cites W2011265495 @default.
• W2019909176 cites W2011561395 @default.
• W2019909176 cites W2015532569 @default.
• W2019909176 cites W2016455305 @default.
• W2019909176 cites W2019371219 @default.
• W2019909176 cites W2020791644 @default.
• W2019909176 cites W2021274488 @default.
• W2019909176 cites W2027838193 @default.
• W2019909176 cites W2031658574 @default.
• W2019909176 cites W2032340480 @default.
• W2019909176 cites W2036691323 @default.
• W2019909176 cites W2039858203 @default.
• W2019909176 cites W2041767379 @default.
• W2019909176 cites W2044289133 @default.
• W2019909176 cites W2044604553 @default.
• W2019909176 cites W2055473086 @default.
• W2019909176 cites W2058635102 @default.
• W2019909176 cites W2059746123 @default.
• W2019909176 cites W2066477421 @default.
• W2019909176 cites W2066931620 @default.
• W2019909176 cites W2098205603 @default.
• W2019909176 cites W2100733433 @default.
• W2019909176 cites W2102990987 @default.
• W2019909176 cites W2104261508 @default.
• W2019909176 cites W2104695435 @default.
• W2019909176 cites W2107886772 @default.
• W2019909176 cites W2109284574 @default.
• W2019909176 cites W2110103060 @default.
• W2019909176 cites W2112191288 @default.
• W2019909176 cites W2112497174 @default.
• W2019909176 cites W2113576610 @default.
• W2019909176 cites W2118125647 @default.
• W2019909176 cites W2136266574 @default.
• W2019909176 cites W2136582516 @default.
• W2019909176 cites W2138895288 @default.
• W2019909176 cites W2144095870 @default.
• W2019909176 cites W2146267034 @default.
• W2019909176 cites W2150194358 @default.
• W2019909176 cites W2150367061 @default.
• W2019909176 cites W2151783777 @default.
• W2019909176 cites W2158130856 @default.
• W2019909176 cites W2158601670 @default.
• W2019909176 cites W2168648043 @default.
• W2019909176 cites W2168815228 @default.
• W2019909176 cites W2169134378 @default.
• W2019909176 cites W2883628765 @default.
• W2019909176 cites W4232180350 @default.
• W2019909176 cites W4241653134 @default.
• W2019909176 cites W4322703283 @default.
• W2019909176 doi "https://doi.org/10.1016/j.neuron.2014.01.031" @default.
• W2019909176 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/24656257" @default.
• W2019909176 hasPublicationYear "2014" @default.
• W2019909176 type Work @default.
• W2019909176 sameAs 2019909176 @default.
• W2019909176 citedByCount "260" @default.
• W2019909176 countsByYear W20199091762014 @default.
• W2019909176 countsByYear W20199091762015 @default.
• W2019909176 countsByYear W20199091762016 @default.
• W2019909176 countsByYear W20199091762017 @default.
• W2019909176 countsByYear W20199091762018 @default.
• W2019909176 countsByYear W20199091762019 @default.
• W2019909176 countsByYear W20199091762020 @default.
• W2019909176 countsByYear W20199091762021 @default.
• W2019909176 countsByYear W20199091762022 @default.
• W2019909176 countsByYear W20199091762023 @default.
• W2019909176 crossrefType "journal-article" @default.
• W2019909176 hasAuthorship W2019909176A5018287398 @default.
• W2019909176 hasAuthorship W2019909176A5089018845 @default.
• W2019909176 hasBestOaLocation W20199091761 @default.
• W2019909176 hasConcept C105702510 @default.
• W2019909176 hasConcept C140530291 @default.
• W2019909176 hasConcept C15744967 @default.
• W2019909176 hasConcept C169760540 @default.
• W2019909176 hasConcept C17744445 @default.
• W2019909176 hasConcept C180747234 @default.
• W2019909176 hasConcept C199539241 @default.
• W2019909176 hasConcept C24998067 @default.
• W2019909176 hasConcept C2776946740 @default.
• W2019909176 hasConcept C2778373776 @default.

• next