Targeted stimulation of motor cortex neural ensembles drives learned movements

Targeted stimulation of motor cortex neural ensembles drives learned movements | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Targeted stimulation of motor cortex neural ensembles drives learned movements An Wu , Qiyu Chen , Bin Yu , Soyoung Chae , Zijing Tan , Assaf Ramot , View ORCID Profile Takaki Komiyama doi: https://doi.org/10.1101/2025.01.06.631504 An Wu 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Qiyu Chen 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Bin Yu 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Soyoung Chae 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Zijing Tan 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Assaf Ramot 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Takaki Komiyama 1 Department of Neurobiology, University of California San Diego , La Jolla, CA, USA 2 Center for Neural Circuits and Behavior, University of California San Diego , La Jolla, CA, USA 3 Department of Neurosciences, University of California San Diego , La Jolla, CA, USA 4 Halıcıoğlu Data Science Institute, University of California San Diego , La Jolla, CA, USA 5 Kavli Institute for Brain and Mind, University of California San Diego , La Jolla, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Takaki Komiyama For correspondence: tkomiyama{at}ucsd.edu Abstract Full Text Info/History Metrics Supplementary material Preview PDF Abstract During the execution of learned motor skills, the neural population in the layer 2/3 (L2/3) of the primary motor cortex (M1) expresses a reproducible spatiotemporal activity pattern. It is debated whether M1 actively participates in generating the activity pattern or it only passively reflects patterned inputs. Furthermore, it is unclear whether this learned activity pattern causally drives the learned movement. We addressed these issues using in vivo two-photon calcium imaging combined with holographic optogenetic stimulation of specific ensembles of M1 L2/3 neurons in mice engaged in a skilled lever-press task. We briefly and synchronously stimulated ∼20 neurons whose activity onset in voluntary trials precedes movement onsets. This stimulation, despite lacking temporal patterns, induced movements that resembled the learned movement, while producing spatiotemporal activity patterns in other M1 neurons not directly stimulated that resembled the activity during the voluntary learned movement. Trial-by-trial variability of optogenetically triggered population activity in the non-target neurons correlated with the variability in the induced movements. These trial-by-trial variabilities were predicted by the initial state of M1 population activity immediately preceding the optogenetic stimulation. Stimulation of the neurons whose activity followed voluntary movement onsets failed to induce the learned movement. Thus, the learned activity pattern in M1 L2/3 can be generated when the M1 network is prepared at the optimal initial state and receives precise triggering inputs, supporting the active role of M1 in learned activity generation. The resulting activity pattern then causally drives the learned movement. Main Text Learning and generating motor skills are core functions of the brain, enabling animals to engage with their environment in a purposeful and adaptive manner. The primary motor cortex (M1) is a critical structure for these functions ( 1 – 3 ). Motor learning induces multiple forms of plasticity in M1, particularly in L2/3 neurons, which undergo extensive synaptic reorganization ( 4 – 7 ). This synaptic plasticity is accompanied by the emergence of a reproducible spatiotemporal pattern in the population activity of M1 L2/3 neurons during the execution of the learned motor skill ( 4 ). Numerous studies have investigated the role of the learned activity pattern in M1 through neural recordings and loss-of-function experiments ( 8 – 11 ). However, it remains an open question how the learned activity pattern is generated. Furthermore, whether the learned activity pattern causally instructs the learned movement pattern is unknown. We addressed these issues by combining in vivo two-photon imaging with two-photon holographic optogenetic stimulation of movement-related neuronal ensembles in M1 L2/3 in mice performing a forelimb motor skill. Holographic stimulation of early-onset neuronal ensemble drives learned movement We used a cued lever-press task, in which water-restricted mice were trained to grasp a lever and, in response to a sound cue delivered after a variable inter-trial interval, push and hold the lever for one second to acquire a water reward ( Fig. 1A ). After two weeks of daily training, mice became experts at this task, collecting rewards in most trials with short latencies ( Fig. 1B, C ). Furthermore, the movement patterns became consistent across trials, a hallmark of learned motor skills ( Fig. 1C ). These features of motor learning are consistent with previous reports using similar tasks ( 4 – 7, 11 – 13 ). It has been shown that the execution of the motor skill at this expert stage requires M1 activity ( 11 ). It was also found that training with these tasks over additional months can make the behavior independent of M1 ( 11 , 14 ) and so in the current study we focused our experiments on the early expert stage (2-3 weeks). Download figure Open in new tab Fig 1. Holographic stimulation of early-onset neurons drives the learned movement. ( A ) Schematic of experimental setup and task structure. ITI: inter-trial interval. ( B ) Example lever trajectories of rewarded trials in an expert session from one mouse. Grey, individual trials; black, mean. ( C ) Reward rate, reaction time, cue onset to reward time, and movement stereotypy (median trial-to-trial correlation coefficients of lever movements for each session) at expert stage (N = 14 mice). ( D ) (Left) Schematic of viral injection into M1 to label L2/3 neurons. (Middle) Example field of view of in vivo two-photon calcium imaging. (Right) Z-scored time series of estimated spikes from example neurons outlined in the image in the middle. ( E ) (Left) Average population activity of all movement-active, movement-suppressed and non-movement-related neurons. Mean ± SEM across session-averages. (Right) Proportions of movement-active, movement-suppressed and non-movement-related neurons. N = 14779 neurons from 27 sessions from 14 mice. ( F ) (Left) Activity of movement-active (top, sorted by activity onset), movement-suppressed (middle), and non-movement-related (bottom) neurons averaged across movements, and aligned to movement onset. Vertical dashed line indicates movement onset. Each row represents one neuron. N = 14779 neurons from 27 sessions from 14 mice. (Right) Activity of example early-onset movement-active neurons averaged across movements, aligned to movement onset. Mean ± SEM. ( G ) Trial structures for holographic stimulation during behavioral sessions. ( H ) (Left) Example two-photon calcium imaging of three planes in M1 L2/3 spanning 60 µm in depth, showing target neurons (red) and non-target neurons (beige). (Right) Trial-averaged responses of early-onset target and non-target neurons in opto trials aligned to stimulation onset. ( I ) Fraction of trials with a movement generated within 0.3 sec from trial onset (voluntary vs opto: p = 2.44×10 −6 , voluntary vs omission: p = 8.74×10 −31 , opto vs omission: p = 8.55×10 −13 , mixed-effects model). Box and whisker plots: median and interquartile range, + indicates mean. N = 17 sessions from 11 mice. ( J ) (Left) Example movements from one session. Left, the learned movement; right, voluntary movements and opto-induced movements of various correlations with the learned movement. ( Right) Probability of movements with varying correlations with the learned movement in voluntary and opto trials, respectively. Dashed lines indicate the medians (voluntary: 0.785, opto: 0.455). N = 17 sessions from 11 mice. ( K ) (Left) Activity of movement-active, movement-suppressed, and non-movement-related neurons (separated by horizontal dashed lines) averaged across learned and opto-induced movements separately, from an example session. Each row represents a non-target neuron. (Right) Average activity of non-target neurons during the 1 sec following cue/stimulation onset sorted by amplitudes in learned movements from the same example session (Pearson’s correlation; p =5.72×10 −17 , R=0.441). ( L ) (Left) Average activity of non-target neurons during voluntary learned movements versus during opto-induced movements pooled across sessions ( p = 1.15×10 −150 , mixed-effects model; Pearson’s correlation, R=0.249). Each dot represents the activity of one neuron averaged during the 1 sec following cue/stimulation onset, averaged across trials. (Right) Average activity of non-target neurons during voluntary learned movements versus during all voluntary movements pooled across sessions ( p < 1.00×10 −308 , mixed-effects model; Pearson’s correlation, R=0.645). Each dot represents the activity of one neuron averaged during the 1 sec following cue onset, averaged across trials. N = 17 sessions from 11 mice. For this and all other figures, *: p <0.05; **: p <0.01; ***: p <0.001. To record and stimulate the M1 L2/3 population, we injected a bicistronic AAV vector to co-express the calcium indicator GCaMP6m and the excitatory opsin ChRmine in the same neurons ( 15 ). We applied in vivo two-photon calcium imaging to record M1 L2/3 activity in task-performing mice ( Fig. 1D ). This approach revealed heterogeneous activity, including neurons showing significantly elevated activity (‘movement-active’, 29.7%) or decreased activity (‘movement-suppressed’, 16.2%) during movements ( Fig. 1E ). The activity onset of individual movement-active neurons relative to movement onset tiled the duration of movements. Of these neurons, we identified a group of ‘early-onset neurons’ (12.8 ± 1.5% of all neurons per session) whose activity onset preceded the movement onset ( Fig. 1F ). We hypothesized that activity in these early-onset neurons contributed to the initiation of the learned movement. To test this hypothesis, we leveraged two-photon holographic optogenetic stimulation with a spatial light modulator that enables the activation of arbitrary groups of neurons within the field of view with single-cell resolution ( 15 – 19 ). We used this approach to ask whether an artificial stimulation of the early-onset neurons could induce the learned movement. We first performed two-photon calcium imaging for ∼60 trials, imaging at 3 depths within M1 L2/3, each spaced 30 µm apart, to capture 547 ± 41 neurons. The data were quickly analyzed offline to identify early-onset neurons, after which the same imaging fields were identified, and the behavioral session was resumed. From this point forward, a majority of the trials were normal voluntary trials in which a lever press after the sound cue induced a water reward. In 30% of randomly interspersed trials (‘opto trials’), the sound cue was omitted, and instead a brief 3 ms pulse of optogenetic stimulation was delivered to ∼20 early-onset neurons simultaneously at the trial onset (see Methods). Critically, in opto trials, no reward was given even when mice made a successful movement. We also included omission trials, in which the sound cue and optogenetic stimulation were omitted, and no reward was ever given ( Fig. 1G ). In the opto trials, the stimulation successfully elicited reliable activity in the target neurons as revealed by simultaneous calcium imaging ( Fig. 1H ). We observed that the stimulation of ∼20 early-onset neurons can trigger movements. We focused our analysis on lever movements initiated within a short latency (0.3 sec) after trial onset (see Methods). Even though the fraction of trials in which movements were initiated was slightly higher in voluntary trials than in opto trials, both fractions were substantially higher than in omission trials. This indicates that movements in opto trials were indeed driven by the stimulation as opposed to voluntary movements serendipitously occurring after the stimulation ( Fig. 1I ). We next examined whether the opto-induced movements resembled the learned movement, which was defined for each session by identifying the subset of the voluntary movements that exhibited high stereotypy and averaging the movements in a random half of those trials. The trials used to construct the learned movement were excluded from subsequent analysis to avoid self-comparisons (see Methods). Both voluntary movements and opto-induced movements in individual trials showed variable correlations with the learned movement, with voluntary movements showing an overall higher correlation. Notably, however, about half of the opto-induced movements showed a high (>0.5) correlation with the learned movement ( Fig. 1J ). Thus, a brief and synchronous stimulation of ∼20 early-onset neurons frequently induced movements, many but not all of which resembled the learned movement. How could 20 neurons drive the learned movement? M1 L2/3 is a highly recurrent network, where excitatory neurons form dense interconnections ( 20 – 23 ). We hypothesized that the activation of these ∼20 neurons elicit population activity in the other, non-target neurons that resembles the activity during the voluntary learned movement. To investigate this, we analyzed the trial-averaged activity of the non-target neurons during the learned movements generated in voluntary trials (‘learned activity’) and compared that to the trial-averaged activity during opto-induced movements. We found that the activity of non-target neurons during opto-induced movements correlated with those during the voluntary learned movements ( Fig. 1K, L ) . In other words, neurons that were activated and suppressed during the voluntary learned movements tended to be also activated and suppressed during opto-induced movements, respectively. To explore the upper bound of this correlation, we repeated this analysis using the activity of voluntary trials (note that not all voluntary trials resulted in the learned movement, Fig. 1J ). The voluntary trial activity correlated with the learned activity, similarly to (but more strongly than) the opto-induced activity ( Fig. 1L ). Thus, our stimulation of ∼20 early-onset neurons induced population activity patterns that resembled the learned pattern in the non-target neurons that are not directly stimulated. Stimulation of late-onset neuronal ensemble fails to drive learned movement It is striking that our artificial stimulation of a small group of neurons induced naturalistic population activity. One possibility is that any random input can engage the local pattern-completion network and induce the same naturalistic activity. To test this, we conducted another set of experiments, this time targeting a subset of ‘late-onset neurons’ (14.2 ± 0.8% of all neurons per session), whose activity onsets were later than movement onsets in voluntary movements ( Fig. 2A, B ). ∼20 of the late-onset neurons were randomly selected as target neurons, matching the number stimulated for the early-target stimulation experiments. The stimulation was done during behavioral sessions in which a majority of trials were voluntary trials, in the same way as early-target stimulation sessions. Download figure Open in new tab Fig 2. Stimulation of late-onset neurons does not drive the learned movement. ( A ) (Left) Activity of movement-active (top, sorted by activity onset), movement-suppressed (middle), and non-movement-related (bottom) neurons averaged across movements, and aligned to movement onset. Vertical dashed line indicates movement onset. Each row represents one neuron. N = 14779 neurons from 27 sessions from 14 mice. (Right) Activity of example late-onset movement-active neurons averaged across movements, aligned to movement onset. ( B ) (Left) Example two-photon calcium imaging of three planes in M1 L2/3 spanning 60 µm in depth, showing target neurons (blue) and non-target neurons (beige). (Right) Trial-averaged responses of late-onset target and non-target neurons in opto trials aligned to stimulation onset. ( C ) Fraction of trials with a movement generated within 0.3 sec from trial onset (voluntary vs early: p = 5.79×10 −6 , voluntary vs late: p = 5.90×10 −13 , voluntary vs omission: p = 7.39×10 −37 , early vs late: p = 6.50×10 −3 , early vs omission: p = 1.00×10 −17 , late vs omission: p = 1.04×10 −8 , mixed-effects model). ( D ) (Left) Probability of movements with varying correlations with the learned movement for each trial type. Dashed lines indicate the medians (voluntary: 0.723, early: 0.455, late: 0.229). (Right) Correlation with the learned movement in each trial type (voluntary vs early: p = 4.28×10 −17 , voluntary vs late: p = 8.26×10 −15 , early vs late: p = 3.78×10 −4 , mixed-effects model). Box and whisker plots: median and interquartile range, + indicates mean. ( E ) Reaction time of movements in each trial type (voluntary vs early: p = 1.39×10 −8 , voluntary vs late: p = 1.66×10 −18 , early vs late: p = 3.62×10 −7 , mixed-effects model). Box and whisker plots: median and interquartile range, + indicates mean. ( F ) (Left) Fraction of responsive non-target neurons per trial for early-target and late-target stimulation trials ( p = 7.10×10 −5 , mixed-effects model). (Right) Response amplitudes of the responsive non-target neurons for early-target and late-target stimulation trials ( p = 0.887, mixed-effects model). ( G ) (Left) Average activity of non-target neurons during voluntary learned movements versus in early-target stimulation trials pooled across sessions ( p = 1.15×10 −150 , mixed-effects model; Pearson’s correlation, R=0.249). Each dot represents the activity of one neuron averaged during the 1 sec following cue/stimulation onset, averaged across trials. (Middle) Average activity of non-target neurons during voluntary learned movements versus in late-target stimulation trials pooled across sessions ( p = 3.70×10 −15 , mixed-effects model; Pearson’s correlation, R=0.111). Each dot represents the activity of one neuron averaged during the 1 sec following cue/stimulation onset, averaged across trials. (Right) Correlation coefficients between non-target neuron population activity in voluntary learned movements and opto-induced movements calculated for individual sessions for early-target and late-target stimulation sessions, respectively ( p = 2.24×10 −4 , Wilcoxon rank-sum test). For this figure, N = 17 early-target stimulation sessions from 11 mice and 10 late-target stimulation sessions from 8 mice with the exception of Fig. 2A . Late-target stimulation induced movements in a significantly lower fraction of trials compared to voluntary trials and early-target stimulation trials, even though the fraction in the late-target trials was higher than in the omission trials ( Fig. 2C ). We next examined the quality of stimulation-induced movements by evaluating the correlation with the learned movement. We found that the movements induced by late-target stimulation were significantly less similar to the learned pattern than the movements in voluntary and early-target stimulation trials ( Fig. 2D ). Thus, compared to early-target stimulation, late-target stimulation induced movements less frequently, and the small number of induced movements tended not to resemble the learned movement. These results indicate that the choice of target neurons is important—early-target stimulation is much more effective at inducing the learned movement than late-target stimulation. We also observed differences in the reaction time of movement initiation across trial types. Specifically, opto-induced movements in both early-target and late-target stimulation trials had a significantly shorter reaction time than voluntary movements ( Fig. 2E ). The shorter reaction time supports the idea that holographic stimulation of M1 L2/3 neurons bypassed the sensory-motor transformation taking place in voluntary trials, leading to faster movement onsets. Interestingly, the reaction time was shorter in late-target stimulation than in early-target stimulation. This raises the possibility that early-target stimulation induced an additional circuit computation to generate the learned activity. We next explored the neural underpinnings that differentiate behavioral outcomes between early-target and late-target stimulation. To this end, we analyzed the activity of the non-target neurons within M1 L2/3 and how they were recruited by the stimulation of the target neurons. We found that early-target stimulation led to the activation of a larger fraction of non-target neurons than late-target stimulation, while the response amplitudes of the recruited non-target neurons were similar in both groups ( Fig. 2F ). Furthermore, we found that the population activity induced by late-target stimulation did not resemble the learned activity as much as early-target stimulation ( Fig. 2G ). Overall, stimulation of late-onset ensembles failed to reliably induce the learned movement, likely because the elicited population activity patterns did not sufficiently match the activity pattern underlying the learned movement. These findings suggest that generating the learned movement requires the activation of early-onset neuronal ensembles as the triggering input to the M1 network, which then propagate the activity to other nearby neurons to complete the learned activity pattern. M1 activity explains movement variability Opto-induced movements exhibited varying degrees of similarity to the learned movement pattern across trials. We next examined whether this variability is reflected in the variability of induced population activity. We focused on the early-target stimulation sessions where we were successful at triggering movements that resembled the learned movement in a considerable fraction of trials. We applied principal component analysis (PCA) on the non-target population activity (both voluntary and opto trials combined), and then projected the population activity of individual trials to the neural space defined by the top 3 PCs ( fig. S1 , see Methods). In voluntary trials, we observed that movements with higher correlations to the learned movement exhibited population activity trajectories closely aligned with those during the learned movement (i.e. the learned activity pattern) ( Fig. 3A ). Additionally, on a trial-by-trial basis, there was a negative correlation between the population activity distance to the learned activity and movement correlation with the learned movement ( Fig. 3A ). In other words, the more similar the population activity was to the learned activity, the more similar the movement was to the learned movement. A similar result was observed in opto trials, where movements with low correlations with the learned movement showed population activity trajectories distinct from the learned activity ( Fig. 3B ). These findings suggest that the learned movement was generated when the induced population activity resembled the learned activity. Download figure Open in new tab Fig 3. M1 activity explains movement variability. ( A ) (Left) Population activity trajectories of all non-target neurons in voluntary trials of an example session, for the 1 second after cue onset. Black, learned activity; grey dotted and solid lines, trials with high (>0.7) and low (≤0.5) correlation with the learned movement respectively. (Right) Population activity distance to the learned activity negatively correlated with correlation with the learned movement ( p = 2.98×10 −8 , mixed-effects model; Pearson’s correlation, R=-0.295). Each dot represents an individual trial. ( B ) Population activity trajectories of all non-target neurons in opto trials of an example session, for the 1 second after stimulation onset. Black, learned activity; pink dotted and solid lines, trials with high (>0.7) and low (≤0.5) correlation to the learned movement respectively. (Right) Population activity distance to the learned activity negatively correlated with correlation with the learned movement ( p = 1.92×10 −7 , mixed-effects model; Pearson’s correlation, R=-0.301). Each dot represents an individual trial. ( C ) Population activity distance negatively correlated with movement correlation between voluntary trials (Left, p = 7.12×10 −77 , mixed-effects model; Pearson’s correlation, R=-0.246), between opto trials (Middle, p = 5.57×10 −13 , mixed-effects model; Pearson’s correlation, R=-0.198), and between voluntary and opto trials (Right, p = 5.25×10 −95 , mixed-effects model; Pearson’s correlation, R=-0.225). ( D ) Schematic illustrating the prediction of lever movements on individual trials based on M1 population activity. A regression model was trained using 80% of voluntary trials and tested on the remaining voluntary trials as well as opto trials (see Methods). ( E ) Examples of predicted lever traces. Predicted lever residuals were added to the mean lever traces to reconstruct the full lever traces. ( F ) Prediction accuracy, measured by correlation coefficients between the actual and predicted lever movements, were significantly better than chance in both voluntary and opto trials (voluntary vs shuffled: p = 1.68×10 −3 , opto vs shuffled: p = 4.72×10 −16 , mixed-effects model), and better in opto trials compared to voluntary trials ( p = 5.20×10 −6 , mixed-effects model). For this figure, N = 17 sessions from 11 mice. In addition to the similarity to the learned movement, a previous study showed that trial-by-trial variability in movement kinematics is represented in M1 L2/3 activity of trained animals ( 4 ). Thus, we next examined the activity-movement relationship in pairs of trials to ask whether the relationship is shared across both the voluntary and opto-induced movements. We performed a similar PCA analysis, but instead of comparing individual trials to the learned pattern, we focused on trial-to-trial comparisons. Specifically, we examined voluntary trial pairs, opto trial pairs, and voluntary-opto trial pairs to assess whether the movement similarity and the population activity similarity correlate with each other across voluntary and opto-induced movements. In all three comparisons, we observed a significant negative correlation between the population activity distance and the movement correlation ( Fig. 3C ). This suggests that M1 L2/3 activity reflects the movement kinematics on a trial-by-trial basis, regardless of whether the movement was generated voluntarily or by our optogenetic stimulation. We further explored the activity-movement relationship using a decoding analysis. We used the principal components of the population activity to decode the movement time series on individual trials using lasso regression. The decoder was trained using 80% of voluntary trials and tested on other trials ( Fig. 3D , see Methods). The decoder was able to decode the voluntary movements on held-out trials substantially better than the chance level defined by trial shuffles ( Fig. 3E , F ). This confirms that M1 L2/3 activity contains information about movement kinematics. When the same decoder trained on voluntary trials was tested on opto trials, we again saw that it was able to decode movements substantially better than chance ( Fig. 3E, F ). This suggests that the mechanism by which M1 L2/3 drives movements was shared between voluntary and opto trials. Surprisingly, the decoder performance was significantly better in opto trials than in held-out voluntary trials, even though the decoder was trained with voluntary trials. This may suggest that voluntary movements are also influenced by additional sources of variability outside of M1 which dilutes the coupling between M1 activity and movement, while opto movements are more purely controlled by M1 activity. Taken together, these findings indicate that the variability in M1 L2/3 activity can explain the variability of both voluntary and opto-induced movements. Initial state of M1 activity predicts movement variability What accounts for the variability in the non-target population activity in opto trials, when the input was the same in every trial (3-ms stimulation of ∼20 early-onset neurons)? Previous studies have found that the variability in the state of motor cortex population activity prior to movement initiation contributes to movement variability ( 24 – 28 ). Thus, we hypothesized that the initial population state before the stimulation influences the subsequent evolution of population activity, thereby contributing to movement variability. To address this hypothesis, we examined the population activity during the period (0.5-sec) immediately prior to our stimulation in opto trials or cue onset in voluntary trials. In opto trials, we observed that the pre-stimulation population activity that was closer to the pre-cue state of the learned activity was associated with subsequent, opto-induced movements more similar to the learned movement ( Fig. 4A ). In other words, the pre-stimulation activity state predicted whether the subsequent stimulation was going to be successful in inducing the learned movement. Similar results were found in voluntary trials, such that the pre-cue population state predicted the quality of the voluntary movements ( Fig. 4B ). Thus, in both the opto and voluntary trials, on a trial-by-trial basis, there appears to be an optimal initial state that facilitated the subsequent generation of the learned movement. This is consistent with the ‘initial condition hypothesis’ which posits that preparatory activity sets the initial state of a dynamical system, which then evolves to generate spatiotemporal activity partially defined by the initial state ( 28 ). Supporting this notion, the pre-stimulation and pre-cue activity state predicted the subsequent population activity during movements on a trial-by-trial basis, in both the opto and voluntary trials respectively ( Fig. 4C ). Perhaps as a consequence, we found a negative correlation between the distance of initial states and movement correlation—in pairs of voluntary trials, pairs of opto trials, and pairs of the two trial types ( Fig. 4D ). Download figure Open in new tab Fig 4. Initial neural state predicts movement variability. ( A ) (Left) Population activity trajectories of all non-target neurons in opto trials of an example session, for the 0.5 second period before stimulation onset. Black, learned activity; pink dotted and solid lines, trials with high (>0.7) and low (≤0.5) correlation to the learned movement respectively. (Right) Population activity distance to the learned activity negatively correlated with correlation with the learned movement ( p = 5.03×10 −4 , mixed-effects model; Pearson’s correlation, R=-0.221). Each dot represents an individual trial. ( B ) (Left) Population activity trajectories of all non-target neurons in voluntary trials of an example session, for the 0.5 second period before cue onset. Black, learned activity; grey dotted and solid lines, trials with high (>0.7) and low (≤0.5) correlation with the learned movement respectively. (Right) Population activity distance to the learned activity negatively correlated with correlation with the learned movement ( p = 1.40×10 −3 , mixed-effects model; Pearson’s correlation, R=-0.173). Each dot represents an individual trial. ( C ) Population activity distance to the learned activity showed a correlation between pre- and post-stimulation periods in opto trials (left, p = 5.46×10 −5 , mixed-effects model; Pearson’s correlation, R=0.363) and between pre- and post-cue periods in voluntary trials (right, p = 7.25×10 −7 , mixed-effects model; Pearson’s correlation, R=0.318). ( D ) Population activity distance before the cue or stimulation negatively correlated with correlation of subsequent movements between voluntary trials (left, p = 1.10×10 −10 , mixed-effects model; Pearson’s correlation, R=-0.112), between opto trials (middle, p = 2.26×10 −7 , mixed-effects model; Pearson’s correlation, R=-0.168), and between voluntary and opto trials (right, p = 1.58×10 −17 , mixed-effects model; Pearson’s correlation, R=-0.097). For this figure, N = 17 sessions from 11 mice. Taken together, the initial state of the M1 L2/3 population predicts the variability in subsequent M1 activity and movements. Discussion In this study, we leveraged holographic stimulation to gain insights into the mechanisms that regulate the learned activity in M1 L2/3 and its role in generating learned movements. We found that brief and synchronous stimulation of approximately 20 early-onset, but not late-onset, movement-related neurons often triggered the learned activity within M1 L2/3 and induced the learned movement. The efficacy of this stimulation depended on the initial state of the M1 L2/3 population activity. Taken together, we propose that the execution of the learned movement is ensured by two conditions: first, the motor cortex prepares the movement by entering the appropriate initial state of the population activity. Second, M1 receives specific inputs from other brain area(s) that activate the early-onset neurons. When both of these two conditions are met, M1 generates the learned activity pattern that reliably drives the learned movement. The importance of the initial state aligns well with the dynamical systems perspective on motor cortex movement control ( 27, 29, 30 ). Previous studies have found that the pre-movement activity state of motor cortex correlates with movement parameters ( 25, 28, 29 ), leading to the postulate that the motor cortex functions as a dynamical machine in which its state at one point in time predicts the next state. Our results provide a direct support for this idea by demonstrating that the network response to our identical stimulus is variable depending on the initial state of the network. However, the initial state does not completely specify the subsequent activity, as the external inputs also affect the evolution of network activity shown by the differences observed in our ‘early-target’ vs. ‘late-target’ stimulation. Future studies could explore whether targeted perturbations can correct suboptimal initial states to enhance movement consistency. By directly stimulating a subset of task-relevant neurons in M1 L2/3, we demonstrated that locally initiated cortical activity can ‘pattern-complete’ to drive learned movements. One possibility is that learning shapes the recurrent connectivity within M1 L2/3, forming a local circuit that can autonomously generate the learned activity pattern. In this possibility, activity may initiate in a group of neurons and then propagates through the network like a chain reaction—reminiscent of models proposed in zebra finch song production ( 31 , 32 ). Computational studies have shown that recurrent neural networks are capable of generating complex activity patterns that underlie movements ( 33 , 34 ). M1 L2/3, as the primary input layer of M1 ( 35 ), is inherently highly recurrent in its connectivity and shows a high degree of learning-related plasticity ( 4 – 7, 20 – 23 ), further supporting this possibility. However, we acknowledge that our results do not rule out the possibility that the pattern-completion mechanism involves a reverberation of activity across distant brain areas such as the basal ganglia and thalamus. The degree of autonomy of M1 in generating the activity pattern in M1 has been debated ( 27, 36 – 38 ). We found that temporally unpatterned inputs to M1, when targeted to the correct population of neurons, can lead to a complete pattern of learned activity. We argue that this observation excludes the possibility that M1 is a purely passive machine that only reflects the temporally dynamic activity pattern provided by inputs from other brain areas. Future research should elucidate the degree and specificity of the circuit reorganization within M1 L2/3 versus upstream areas, and the relative contributions of local vs. global interactions in cortical pattern generation. Author contributions Conceptualization: AW, TK Methodology: AW, QC, TK Investigation: AW, QC, BY, ZT Visualization: QC Formal analysis: AW, QC, SC Validation: QC Resources: AW, QC, TK Software: AW, QC Data curation: QC Funding acquisition: TK Project administration: TK Supervision: TK Writing – original draft: QC, TK Writing – review & editing: QC, BY, SC, ZT, AR, TK Competing interests The authors declare they have no competing interests. Acknowledgments We cherish our memories of our dear friend and colleague, An Wu, and dedicate this manuscript to her surviving parents, Mr. Wu and Ms. Qu. We thank Bobbie Morales, Ashley Medina, David Arakelyan, and Elanore Hall for technical assistance; and other Komiyama Lab members, especially J. Li and E. Gjoni, for discussion. We also thank all our friends who have reached out to us to offer support during the difficult time. Footnotes ↵ # Deceased in a tragic fire on March 16, 2023 References 1. H. Makino , E. J. Hwang , N. G. Hedrick , T. Komiyama , Circuit Mechanisms of Sensorimotor Learning . Neuron 92 , 705 – 721 ( 2016 ). OpenUrl CrossRef PubMed 2. A. J. Peters , H. Liu , T. Komiyama , Learning in the Rodent Motor Cortex . Annual Review of Neuroscience 40 , 77 – 97 ( 2017 ). OpenUrl CrossRef PubMed 3. E. Kogan , J. Lu , Y. Zuo , Cortical circuit dynamics underlying motor skill learning: from rodents to humans . Front. Mol. Neurosci . 16 ( 2023 ). 4. ↵ A. J. Peters , S. X. Chen , T. Komiyama , Emergence of reproducible spatiotemporal activity during motor learning . Nature 510 , 263 – 267 ( 2014 ). OpenUrl CrossRef PubMed Web of Science 5. S. X. Chen , A. N. Kim , A. J. Peters , T. Komiyama , Subtype-specific plasticity of inhibitory circuits in motor cortex during motor learning . Nature Neuroscience 18 , 1109 – 1115 ( 2015 ). OpenUrl CrossRef PubMed 6. N. G. Hedrick , Z. Lu , E. Bushong , S. Singhi , P. Nguyen , Y. Magaña , S. Jilani , B. K. Lim , M. Ellisman , T. Komiyama , Learning binds new inputs into functional synaptic clusters via spinogenesis . Nat Neurosci 25 , 726 – 737 ( 2022 ). OpenUrl CrossRef PubMed 7. N. G. Hedrick , W. J. Wright , T. Komiyama , Local and global predictors of synapse elimination during motor learning . Science Advances 10 , eadk0540 ( 2024 ). OpenUrl CrossRef PubMed 8. L. Guo , H. Xiong , J. I. Kim , Y. W. Wu , R. R. Lalchandani , Y. Cui , Y. Shu , T. Xu , J. B. Ding , Dynamic rewiring of neural circuits in the motor cortex in mouse models of Parkinson’s disease . Nature Neuroscience 18 , 1299 – 1309 ( 2015 ). OpenUrl CrossRef PubMed 9. R. Kawai , T. Markman , R. Poddar , R. Ko , A. L. Fantana , A. K. Dhawale , A. R. Kampff , B. P. Ölveczky , Motor Cortex Is Required for Learning but Not for Executing a Motor Skill . Neuron 86 , 800 – 812 ( 2015 ). OpenUrl CrossRef PubMed 10. J.-Z. Guo , A. R. Graves , W. W. Guo , J. Zheng , A. Lee , J. Rodríguez-González , N. Li , J. J. Macklin , J. W. Phillips , B. D. Mensh , K. Branson , A. W. Hantman , Cortex commands the performance of skilled movement . eLife 4 , e10774 ( 2015 ). OpenUrl CrossRef PubMed 11. ↵ E. J. Hwang , J. E. Dahlen , Y. Y. Hu , K. Aguilar , B. Yu , M. Mukundan , A. Mitani , T. Komiyama , Disengagement of motor cortex from movement control during long-term learning . Science Advances 5 , 1 – 13 ( 2019 ). OpenUrl CrossRef PubMed 12. C. Ren , K. Peng , R. Yang , W. Liu , C. Liu , T. Komiyama , Global and subtype-specific modulation of cortical inhibitory neurons regulated by acetylcholine during motor learning . Neuron 110 , 2334 - 2350 .e8 ( 2022 ). OpenUrl CrossRef PubMed 13. H. Makino , C. Ren , H. Liu , A. N. Kim , N. Kondapaneni , X. Liu , D. Kuzum , T. Komiyama , Transformation of Cortex-wide Emergent Properties during Motor Learning . Neuron 94 , 880 - 890 .e8 ( 2017 ). OpenUrl CrossRef PubMed 14. ↵ E. J. Hwang , J. E. Dahlen , M. Mukundan , T. Komiyama , Disengagement of Motor Cortex during Long-Term Learning Tracks the Performance Level of Learned Movements . J. Neurosci . 41 , 7029 – 7047 ( 2021 ). OpenUrl Abstract / FREE Full Text 15. ↵ J. H. Marshel , Y. S. Kim , T. A. Machado , S. Quirin , B. Benson , J. Kadmon , C. Raja , A. Chibukhchyan , C. Ramakrishnan , M. Inoue , J. C. Shane , D. J. McKnight , S. Yoshizawa , H. E. Kato , S. Ganguli , K. Deisseroth , Cortical layer–specific critical dynamics triggering perception . Science 365 ( 2019 ). 16. A. M. Packer , L. E. Russell , H. W. P. Dalgleish , M. Häusser , Simultaneous all-optical manipulation and recording of neural circuit activity with cellular resolution in vivo . Nat Methods 12 , 140 – 146 ( 2015 ). OpenUrl CrossRef PubMed 17. H. Adesnik , L. Abdeladim , Probing neural codes with two-photon holographic optogenetics . Nat Neurosci 24 , 1356 – 1366 ( 2021 ). OpenUrl CrossRef PubMed 18. L. E. Russell , H. W. P. Dalgleish , R. Nutbrown , O. M. Gauld , D. Herrmann , M. Fişek , A. M. Packer , M. Häusser , All-optical interrogation of neural circuits in behaving mice . Nature Protocols , doi: 10.1038/s41596-022-00691-w ( 2022 ). OpenUrl CrossRef PubMed 19. S. C. Piantadosi , Z. C. Zhou , C. Pizzano , C. E. Pedersen , T. K. Nguyen , S. Thai , G. D. Stuber , M. R. Bruchas , Holographic stimulation of opposing amygdala ensembles bidirectionally modulates valence-specific behavior via mutual inhibition . Neuron , doi: 10.1016/j.neuron.2023.11.007 ( 2023 ). OpenUrl CrossRef 20. R. J. Douglas , K. A. C. Martin , NEURONAL CIRCUITS OF THE NEOCORTEX . Annual Review of Neuroscience 27 , 419 – 451 ( 2004 ). OpenUrl CrossRef PubMed Web of Science 21. R. J. Douglas , K. A. C. Martin , Recurrent neuronal circuits in the neocortex . Current Biology 17 , R496 – R500 ( 2007 ). OpenUrl CrossRef PubMed Web of Science 22. K. D. Harris , G. M. G. Shepherd , The neocortical circuit: themes and variations . Nature Neuroscience 18 , 170 – 181 ( 2015 ). OpenUrl CrossRef PubMed 23. R. Muñoz-Castañeda , B. Zingg , K. S. Matho , X. Chen , Q. Wang , N. N. Foster , A. Li , A. Narasimhan , K. E. Hirokawa , B. Huo , S. Bannerjee , L. Korobkova , C. S. Park , Y.-G. Park , M. S. Bienkowski , U. Chon , D. W. Wheeler , X. Li , Y. Wang , M. Naeemi , P. Xie , L. Liu , K. Kelly , X. An , S. M. Attili , I. Bowman , A. Bludova , A. Cetin , L. Ding , R. Drewes , F. D’Orazi , C. Elowsky , S. Fischer , W. Galbavy , L. Gao , J. Gillis , P. A. Groblewski , L. Gou , J. D. Hahn , J. T. Hatfield , H. Hintiryan , J. J. Huang , H. Kondo , X. Kuang , P. Lesnar , X. Li , Y. Li , M. Lin , D. Lo , J. Mizrachi , S. Mok , P. R. Nicovich , R. Palaniswamy , J. Palmer , X. Qi , E. Shen , Y.-C. Sun , H. W. Tao , W. Wakemen , Y. Wang , S. Yao , J. Yuan , H. Zhan , M. Zhu , L. Ng , L. I. Zhang , B. K. Lim , M. Hawrylycz , H. Gong , J. C. Gee , Y. Kim , K. Chung , X. W. Yang , H. Peng , Q. Luo , P. P. Mitra , A. M. Zador , H. Zeng , G. A. Ascoli , Z. Josh Huang , P. Osten , J. A. Harris , H.-W. Dong , Cellular anatomy of the mouse primary motor cortex . Nature 598 , 159 – 166 ( 2021 ). OpenUrl CrossRef PubMed 24. A. Riehle , J. Requin , The predictive value for performance speed of preparatory changes in neuronal activity of the monkey motor and premotor cortex . Behav Brain Res 53 , 35 – 49 ( 1993 ). OpenUrl CrossRef PubMed Web of Science 25. M. M. Churchland , A. Afshar , K. V. Shenoy , A Central Source of Movement Variability . Neuron 52 , 1085 – 1096 ( 2006 ). OpenUrl CrossRef PubMed Web of Science 26. M. M. Churchland , J. P. Cunningham , M. T. Kaufman , S. I. Ryu , K. V. Shenoy , Cortical Preparatory Activity: Representation of Movement or First Cog in a Dynamical Machine? Neuron 68 , 387 – 400 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 27. K. V. Shenoy , M. Sahani , M. M. Churchland , Cortical Control of Arm Movements: A Dynamical Systems Perspective . Annual Review of Neuroscience 36 , 337 – 359 ( 2013 ). OpenUrl CrossRef PubMed Web of Science 28. ↵ A. Afshar , G. Santhanam , B. M. Yu , S. I. Ryu , M. Sahani , K. V. Shenoy , Single-Trial Neural Correlates of Arm Movement Preparation . Neuron 71 , 555 – 564 ( 2011 ). OpenUrl CrossRef PubMed Web of Science 29. M. M. Churchland , K. V. Shenoy , Preparatory activity and the expansive null-space . Nat. Rev. Neurosci . 25 , 213 – 236 ( 2024 ). OpenUrl CrossRef PubMed 30. S. Vyas , M. D. Golub , D. Sussillo , K. V. Shenoy , Computation Through Neural Population Dynamics . Annual Review of Neuroscience 43 , 249 – 275 ( 2020 ). OpenUrl CrossRef PubMed 31. ↵ M. A. Long , D. Z. Jin , M. S. Fee , Support for a synaptic chain model of neuronal sequence generation . Nature 468 , 394 – 399 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 32. ↵ F. W. Moll , D. Kranz , A. Corredera Asensio , M. Elmaleh , L. A. Ackert-Smith , M. A. Long , Thalamus drives vocal onsets in the zebra finch courtship song . Nature 616 , 132 – 136 ( 2023 ). OpenUrl CrossRef PubMed 33. ↵ R. Laje , D. V. Buonomano , Robust timing and motor patterns by taming chaos in recurrent neural networks . Nature Neuroscience 16 , 925 – 933 ( 2013 ). OpenUrl CrossRef PubMed 34. ↵ L. Bachschmid-Romano , N. G. Hatsopoulos , N. Brunel , Interplay between external inputs and recurrent dynamics during movement preparation and execution in a network model of motor cortex . eLife 12 , e77690 ( 2023 ). OpenUrl CrossRef PubMed 35. ↵ N. Weiler , L. Wood , J. Yu , S. A. Solla , G. M. G. Shepherd , Top-down laminar organization of the excitatory network in motor cortex . Nature Neuroscience 11 , 360 – 366 ( 2008 ). OpenUrl CrossRef PubMed Web of Science 36. B. A. Sauerbrei , J.-Z. Guo , J. D. Cohen , M. Mischiati , W. Guo , M. Kabra , N. Verma , B. Mensh , K. Branson , A. W. Hantman , Cortical pattern generation during dexterous movement is input-driven . Nature 577 , 386 – 391 ( 2020 ). OpenUrl CrossRef PubMed 37. R. Yuste , J. N. MacLean , J. Smith , A. Lansner , The cortex as a central pattern generator . Nat Rev Neurosci 6 , 477 – 483 ( 2005 ). OpenUrl CrossRef PubMed Web of Science 38. C. Pandarinath , K. C. Ames , A. A. Russo , A. Farshchian , L. E. Miller , E. L. Dyer , J. C. Kao , Latent Factors and Dynamics in Motor Cortex and Their Application to Brain–Machine Interfaces . J. Neurosci . 38 , 9390 – 9401 ( 2018 ). OpenUrl Abstract / FREE Full Text 39. M. Pachitariu , C. Stringer , M. Dipoppa , S. Schröder , L. F. Rossi , H. Dalgleish , M. Carandini , K. D. Harris , Suite2p: beyond 10,000 neurons with standard two-photon microscopy . bioRxiv [Preprint] ( 2017 ). doi: 10.1101/061507 . OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted January 06, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Targeted stimulation of motor cortex neural ensembles drives learned movements Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Targeted stimulation of motor cortex neural ensembles drives learned movements An Wu , Qiyu Chen , Bin Yu , Soyoung Chae , Zijing Tan , Assaf Ramot , Takaki Komiyama bioRxiv 2025.01.06.631504; doi: https://doi.org/10.1101/2025.01.06.631504 Share This Article: Copy Citation Tools Targeted stimulation of motor cortex neural ensembles drives learned movements An Wu , Qiyu Chen , Bin Yu , Soyoung Chae , Zijing Tan , Assaf Ramot , Takaki Komiyama bioRxiv 2025.01.06.631504; doi: https://doi.org/10.1101/2025.01.06.631504 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Neuroscience Subject Areas All Articles Animal Behavior and Cognition (7637) Biochemistry (17705) Bioengineering (13899) Bioinformatics (41968) Biophysics (21460) Cancer Biology (18603) Cell Biology (25526) Clinical Trials (138) Developmental Biology (13385) Ecology (19910) Epidemiology (2067) Evolutionary Biology (24328) Genetics (15614) Genomics (22513) Immunology (17741) Microbiology (40423) Molecular Biology (17193) Neuroscience (88646) Paleontology (667) Pathology (2835) Pharmacology and Toxicology (4827) Physiology (7647) Plant Biology (15160) Scientific Communication and Education (2046) Synthetic Biology (4302) Systems Biology (9825) Zoology (2271)