FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2771907952> ?p ?o ?g. }

• W2771907952 endingPage "83.e4" @default.
• W2771907952 startingPage "77" @default.
• W2771907952 abstract "•Mice can learn to modulate activity difference of two inhibitory neurons•Mice employ subtype-specific strategies to modulate each neuron•When PV neurons are targeted, activity of negative target decreases•When SOM and VIP neurons are targeted, activity of positive target increases Brain-computer interfaces have seen an increase in popularity due to their potential for direct neuroprosthetic applications for amputees and disabled individuals. Supporting this promise, animals—including humans—can learn even arbitrary mapping between the activity of cortical neurons and movement of prosthetic devices [1Moritz C.T. Perlmutter S.I. Fetz E.E. Direct control of paralysed muscles by cortical neurons.Nature. 2008; 456: 639-642Crossref PubMed Scopus (439) Google Scholar, 2Jarosiewicz B. Chase S.M. Fraser G.W. Velliste M. Kass R.E. Schwartz A.B. Functional network reorganization during learning in a brain-computer interface paradigm.Proc. Natl. Acad. Sci. USA. 2008; 105: 19486-19491Crossref PubMed Scopus (209) Google Scholar, 3Ganguly K. Carmena J.M. Emergence of a stable cortical map for neuroprosthetic control.PLoS Biol. 2009; 7: e1000153Crossref PubMed Scopus (399) Google Scholar, 4Sadtler P.T. Quick K.M. Golub M.D. Chase S.M. Ryu S.I. Tyler-Kabara E.C. Yu B.M. Batista A.P. Neural constraints on learning.Nature. 2014; 512: 423-426Crossref PubMed Scopus (325) Google Scholar]. However, the performance of neuroprosthetic device control has been nowhere near that of limb control in healthy individuals, presenting a dire need to improve the performance. One potential limitation is the fact that previous work has not distinguished diverse cell types in the neocortex, even though different cell types possess distinct functions in cortical computations [5Chen S.X. Kim A.N. Peters A.J. Komiyama T. Subtype-specific plasticity of inhibitory circuits in motor cortex during motor learning.Nat. Neurosci. 2015; 18: 1109-1115Crossref PubMed Scopus (181) Google Scholar, 6Pfeffer C.K. Xue M. He M. Huang Z.J. Scanziani M. Inhibition of inhibition in visual cortex: the logic of connections between molecularly distinct interneurons.Nat. Neurosci. 2013; 16: 1068-1076Crossref PubMed Scopus (774) Google Scholar, 7Pi H.-J. Hangya B. Kvitsiani D. Sanders J.I. Huang Z.J. Kepecs A. Cortical interneurons that specialize in disinhibitory control.Nature. 2013; 503: 521-524Crossref PubMed Scopus (665) Google Scholar] and likely distinct capacities to control brain-computer interfaces. Here, we made a first step in addressing this issue by tracking the plastic changes of three major types of cortical inhibitory neurons (INs) during a neuron-pair operant conditioning task using two-photon imaging of IN subtypes expressing GCaMP6f. Mice were rewarded when the activity of the positive target neuron (N+) exceeded that of the negative target neuron (N−) beyond a set threshold. Mice improved performance with all subtypes, but the strategies were subtype specific. When parvalbumin (PV)-expressing INs were targeted, the activity of N− decreased. However, targeting of somatostatin (SOM)- and vasoactive intestinal peptide (VIP)-expressing INs led to an increase of the N+ activity. These results demonstrate that INs can be individually modulated in a subtype-specific manner and highlight the versatility of neural circuits in adapting to new demands by using cell-type-specific strategies. Brain-computer interfaces have seen an increase in popularity due to their potential for direct neuroprosthetic applications for amputees and disabled individuals. Supporting this promise, animals—including humans—can learn even arbitrary mapping between the activity of cortical neurons and movement of prosthetic devices [1Moritz C.T. Perlmutter S.I. Fetz E.E. Direct control of paralysed muscles by cortical neurons.Nature. 2008; 456: 639-642Crossref PubMed Scopus (439) Google Scholar, 2Jarosiewicz B. Chase S.M. Fraser G.W. Velliste M. Kass R.E. Schwartz A.B. Functional network reorganization during learning in a brain-computer interface paradigm.Proc. Natl. Acad. Sci. USA. 2008; 105: 19486-19491Crossref PubMed Scopus (209) Google Scholar, 3Ganguly K. Carmena J.M. Emergence of a stable cortical map for neuroprosthetic control.PLoS Biol. 2009; 7: e1000153Crossref PubMed Scopus (399) Google Scholar, 4Sadtler P.T. Quick K.M. Golub M.D. Chase S.M. Ryu S.I. Tyler-Kabara E.C. Yu B.M. Batista A.P. Neural constraints on learning.Nature. 2014; 512: 423-426Crossref PubMed Scopus (325) Google Scholar]. However, the performance of neuroprosthetic device control has been nowhere near that of limb control in healthy individuals, presenting a dire need to improve the performance. One potential limitation is the fact that previous work has not distinguished diverse cell types in the neocortex, even though different cell types possess distinct functions in cortical computations [5Chen S.X. Kim A.N. Peters A.J. Komiyama T. Subtype-specific plasticity of inhibitory circuits in motor cortex during motor learning.Nat. Neurosci. 2015; 18: 1109-1115Crossref PubMed Scopus (181) Google Scholar, 6Pfeffer C.K. Xue M. He M. Huang Z.J. Scanziani M. Inhibition of inhibition in visual cortex: the logic of connections between molecularly distinct interneurons.Nat. Neurosci. 2013; 16: 1068-1076Crossref PubMed Scopus (774) Google Scholar, 7Pi H.-J. Hangya B. Kvitsiani D. Sanders J.I. Huang Z.J. Kepecs A. Cortical interneurons that specialize in disinhibitory control.Nature. 2013; 503: 521-524Crossref PubMed Scopus (665) Google Scholar] and likely distinct capacities to control brain-computer interfaces. Here, we made a first step in addressing this issue by tracking the plastic changes of three major types of cortical inhibitory neurons (INs) during a neuron-pair operant conditioning task using two-photon imaging of IN subtypes expressing GCaMP6f. Mice were rewarded when the activity of the positive target neuron (N+) exceeded that of the negative target neuron (N−) beyond a set threshold. Mice improved performance with all subtypes, but the strategies were subtype specific. When parvalbumin (PV)-expressing INs were targeted, the activity of N− decreased. However, targeting of somatostatin (SOM)- and vasoactive intestinal peptide (VIP)-expressing INs led to an increase of the N+ activity. These results demonstrate that INs can be individually modulated in a subtype-specific manner and highlight the versatility of neural circuits in adapting to new demands by using cell-type-specific strategies. Water-restricted mice expressing GCaMP6f in parvalbumin (PV)-, somatostatin (SOM)-, or vasoactive intestinal peptide (VIP)-expressing inhibitory neurons (INs) were trained in a neuron-pair operant conditioning task with two-photon calcium imaging (modified from [8Clancy K.B. Koralek A.C. Costa R.M. Feldman D.E. Carmena J.M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning.Nat. Neurosci. 2014; 17: 807-809Crossref PubMed Scopus (89) Google Scholar]). Briefly, two neurons in layer 2/3 of the primary motor cortex were randomly selected out of those that met predetermined activity criteria (STAR Methods) and designated as positive target neuron (N+) and negative target neuron (N−). In some of the imaging fields (14/36), there were only two labeled neurons that met the activity criteria. The distance between N+ and N− was 191 ± 119 μm (mean ± SD; n = 36 pairs). The calcium signal of the targeted neurons was not saturated (Figures S1A–S1C). Mice were rewarded when the calcium activity of N+ exceeded that of N− by a set threshold (Figure 1A). The reward contingency based on the difference between N+ and N− activity ensured that mice could not solve the task by simply activating all neurons in the area simultaneously. During the trial, the difference of the calcium signal of the two neurons was transformed to create a dynamically frequency-modulated auditory feedback. After each reward, the activity of the targeted neurons had to return to baseline, which resumed the auditory feedback and initiated the next trial. The same neurons were targeted over 4–6 sessions, one session per day, with the same reward contingency. We note that a previous study with a similar approach that targeted neurons of unidentified cell types showed that auditory feedback was essential for the learning of the task [8Clancy K.B. Koralek A.C. Costa R.M. Feldman D.E. Carmena J.M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning.Nat. Neurosci. 2014; 17: 807-809Crossref PubMed Scopus (89) Google Scholar]. However, in the current study, we did not explicitly test the necessity of the auditory feedback. For each of the three major IN types, mice were able to improve the performance over sessions, significantly increasing the reward frequency (PV-INs, p = 0.011, n = 48 sessions, 10 imaging fields in 5 mice, using fitlme in MATLAB hereafter unless otherwise stated; SOM-INs, p = 0.001, 50 sessions, 10 imaging fields in 5 mice; VIP-INs, p = 0.017, 82 sessions, 16 imaging fields in 8 mice; Figures 1B–1D). Immunostaining showed a high degree of overlap between GCaMP6f-expressing and PV-expressing neurons in PV animals and little overlap between GCaMP6f-expressing and PV-expressing neurons in SOM and VIP animals (Figures 1E–1G). In a subset of the animals, we imaged the same neurons for an additional 1–3 contingency degradation sessions, in which rewards were provided without regard to the activity of the targeted neurons. In these sessions, the targeted neurons reached the reward threshold significantly less frequently (the difference between the contingency degradation sessions and the last two training sessions, 2.49 ± 0.64/min [estimate ± SE; p < 0.001; n = 12 and 14 sessions; 7 imaging fields in 4 mice], 3.07 ± 0.71/min [p < 0.001; n = 20 and 20 sessions; 10 imaging fields in 5 mice], and 1.08 ± 0.38/min [p = 0.006; n = 28 and 30 sessions; 15 imaging fields in 7 mice] in PV, SOM, and VIP animals, respectively). These results indicate that mice are indeed able to modulate the activity of IN subtypes. We considered three possible strategies by which mice could achieve an increase in reward frequency (Figure 2A). For example, the reward frequency could increase with an increase in the activity frequency of N+ (Figure 2A; “N+ increase”). Alternatively, a decrease in the activity of N− (Figure 2A; “N− decrease”) or decoupling of activity between N+ and N− (Figure 2A; “decoupling”) can improve the reward frequency by increasing the chance that activity in N+ leads to a reward. In the following analysis, we focused on the frequency of calcium events because the amplitude of the events did not significantly change across sessions (Figures S1D–S1F). When PV-INs were targeted (Figure 2B), the frequency of N+ calcium events did not change significantly, arguing against the N+ increase strategy (Figure 2C). In contrast, the frequency of N− calcium events decreased (p = 0.002), supporting the N− decrease strategy (Figure 2D). The slopes of the changes of N+ activity (Figure 2C) and N− activity (Figure 2D) were significantly different (p = 0.001), indicating the specificity of the decrease of N− activity. Accordingly, the frequency of the N+ calcium events that did not lead to a reward decreased (p = 0.025; Figure 2E). The frequency of N− events during the time periods when N+ was inactive also significantly decreased over sessions (p = 0.011; Figure 2F). This result argues against the decoupling strategy, which would predict that N− events that originally coincided with N+ activity would move into the periods of N+ inactivity. Furthermore, the correlation coefficient between N+ and N− activity during task period did not decrease across sessions (slope estimate ± SE = 0.0145 ± 0.0089/session; p = 0.112). To test whether the changes of neural activity contribute to the improvement in task performance, we conducted mediation analysis [9Falk C.F. Biesanz J.C. Two cross-platform programs for inferences and interval estimation about indirect effects in mediational models.SAGE Open. 2016; 6 (2158244015625445)Crossref Scopus (74) Google Scholar]. In this analysis, we found that N+ activity and N− activity were positively and negatively correlated with task performance, respectively (coefficients of a linear model: 0.57 ± 0.06 [p < 0.001] and −0.22 ± 0.06 [p < 0.001; estimate ± SE]), and there was a significant mediation effect with N− activity decrease (p < 0.001; STAR Methods) and not with N+ activity (p = 0.233). These results indicate that, when PV-INs were targeted, mice specifically decreased the activity of N− while maintaining N+ activity. The task improvement in mice with SOM-IN targeting (Figure 3A) involved a different strategy. In contrast to PV-INs, N+ event frequency increased in the later sessions in SOM-INs (p = 0.003; Figure 3B). This suggests that the N+ increase strategy was utilized to perform the task. Conversely, the frequency of N− events did not change (Figure 3C), nor did the frequency of N+ events that were not associated with rewards (Figure 3D), arguing against the N− decrease hypothesis. The slopes of the activity changes of N+ (Figure 3B) and N− (Figure 3C) were significantly different (p = 0.010). Furthermore, the frequency of N− events during the periods of N+ inactivity did not increase (Figure 3E) and the correlation coefficient between N+ and N− activity during task period did not significantly decrease across sessions (slope estimate ± SE = −0.0009 ± 0.0119/session; p = 0.939), discounting the decoupling strategy. In the mediation analysis, N+ and N− activities were positively and negatively correlated with task performance, respectively (coefficients of a linear model: 0.60 ± 0.03 [p < 0.001] and −0.06 ± 0.02 [p = 0.018; estimate ± SE]), and the mediation effect through N+ activity was significant (p = 0.003) and not through N− activity (p = 0.082). We conclude that mice improved the task performance with SOM-INs primarily by activating specifically the N+ neuron. Next, we investigated VIP-INs (Figure 3F). As with SOM-INs, in VIP mice, the event frequency of N+ significantly increased (p = 0.044; Figure 3G). Neither the frequency of N− events nor the N+ event frequency not associated with reward changed significantly (Figures 3H and 3I), arguing against the N− decrease strategy, although the difference in the slopes of activity changes of N+ (Figure 3G) and N− (Figure 3H) did not reach statistical significance (p = 0.101). The N− event frequency within the periods of N+ inactivity did not change (Figure 3J), and the correlation coefficient between N+ and N− activity during task period did not change (slope estimate ± SE = −0.0103 ± 0.0091/session; p = 0.261), excluding the decoupling strategy. N+ and N− activity were positively and negatively correlated with task performance, respectively (coefficients of a linear model: 0.70 ± 0.03 [p < 0.001] and −0.10 ± 0.04 [p = 0.011; estimate ± SE]), and the mediation effect through N+ activity was significant (p = 0.040). These data demonstrate that VIP mice improved the task performance by increasing N+ activity. To test whether the difference between cell types was significant, we examined whether cell type had a significant effect on the slope of the linear model (STAR Methods; Figures 4A–4D). The reward frequency increase was not different among three cell types. The frequency increase of N+ peaks was significantly larger in SOM neurons than in PV and VIP neurons. The decrease of N− peak frequency was greater in PV neurons than in VIP neurons, and the decrease of the frequency of N+ peaks without rewards was specific to PV. For SOM and VIP mice, we further investigated whether N+/N− activity changed differently from the activity of non-target neurons, which met the same activity criteria. We did not perform this analysis for PV mice because 6 out of 10 imaging fields did not have any non-target neuron that met the activity criteria and the other 4 fields only had 1 non-target neuron. The result shows that N+ activity increase in SOM animals was specific to N+ and significantly greater than the non-target neurons (slope difference from non-target neurons: 1.81 ± 0.73 [p = 0.014; SOM; N+; Figure 4E], −0.68 ± 0.77 [p = 0.376; SOM; N−; Figure 4F], 0.43 ± 0.41 [p = 0.304; VIP; N+; Figure 4E] and −0.12 ± 0.42 [p = 0.770; VIP; N−; Figure 4F; estimate ± SE]). In addition, we simulated reward frequency increase as if each non-target neuron were used in the task as either N+ or N−, and the actual target neuron was used for the other target (for example, in a simulation in which we used a non-target neuron as N+, reward frequency was simulated using that non-target neuron as N+ and the actual N−). Figure 4G shows that the simulated reward frequency increase with non-target neurons used as N+ is significantly lower than the actual reward frequency increase in SOM neurons, but not in VIP neurons. The reward frequency increase was not affected in either SOM or VIP animals when non-target neurons were used as N− (Figure 4H). The results show that the activity increase in SOM animals was specific to N+, leading to improved task performance. To our knowledge, this is the first study to test the plasticity of individual neurons of molecularly identified cell types in a brain-computer interface task. We demonstrate that cell type has a profound impact on the way by which performance improvement is achieved. For example, the activity of a SOM-IN could be increased without activating a second SOM-IN, similarly to a previous study that did not identify the cell type of the targeted neurons (and thus most of the targeted neurons were presumably excitatory) [8Clancy K.B. Koralek A.C. Costa R.M. Feldman D.E. Carmena J.M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning.Nat. Neurosci. 2014; 17: 807-809Crossref PubMed Scopus (89) Google Scholar]. However, we found no evidence that the activity of a PV-IN could be increased without also activating a second PV-IN. Instead, the activity of a PV-IN could be reduced without inactivating a second PV-IN. The performance of brain-machine interfaces may improve in the future if such cell-type-specific constraints on plasticity are considered [10Shenoy K.V. Carmena J.M. Combining decoder design and neural adaptation in brain-machine interfaces.Neuron. 2014; 84: 665-680Abstract Full Text Full Text PDF PubMed Scopus (105) Google Scholar]. The differences in baseline activity levels may have partially contributed to the difference in strategies among subtypes. For example, if the baseline activity level of PV neurons is higher, it might be more difficult to increase N+ activity in PV-INs than in other subtypes. In addition, due to different calcium buffering in each cell, the relationship between spikes and GCaMP6f signals may be different from cell to cell, leaving the possibility for PV neurons to require more spikes to cause a calcium event. Alternatively, the cell-type-specific strategies may be partially explained by the differences in the levels of activity correlation within each cell type (pairwise correlation coefficients of pre-task activity between candidate neurons were PV, 0.61 ± 0.84 [npair = 18]; SOM, 0.14 ± 0.03 [npair = 40]; and VIP, 0.31 ± 0.01 [npair = 394]; mean ± SEM; p < 0.001 for all pairwise comparisons after removing the effect of event rate; Figure S2). Nevertheless, we argue that the difference in correlation is an important reflection of their intrinsic properties. The ability to improve task performance with PV-INs is particularly striking, given the high activity correlation between PV-INs. It has been shown that neural feedback tasks based on the difference of two neural ensembles is harder to learn if the activity of the two ensembles is correlated [8Clancy K.B. Koralek A.C. Costa R.M. Feldman D.E. Carmena J.M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning.Nat. Neurosci. 2014; 17: 807-809Crossref PubMed Scopus (89) Google Scholar]. However, the animals could perform the task with a specific strategy. What could be the potential mechanisms underlying cell-type specificity of strategies? At the cellular level, the performance of this task could be mediated by specific plasticity of N+ and/or N− neurons, such as plasticity of intrinsic excitability and synaptic plasticity of inhibitory and excitatory synapses onto these neurons. There are likely differences among cell types in their ability for these plasticity mechanisms (reviewed in [11Kullmann D.M. Moreau A.W. Bakiri Y. Nicholson E. Plasticity of inhibition.Neuron. 2012; 75: 951-962Abstract Full Text Full Text PDF PubMed Scopus (155) Google Scholar]). For example, repetitive correlated spiking induced spike-timing-dependent plasticity (STDP) in low-threshold spiking (LTS) interneurons (putative non-PV INs) [12Yavorska I. Wehr M. Somatostatin-expressing inhibitory interneurons in cortical circuits.Front. Neural Circuits. 2016; 10: 76Crossref PubMed Scopus (104) Google Scholar], whereas it induced long-term depression in fast spiking interneurons (putative PV-INs) [12Yavorska I. Wehr M. Somatostatin-expressing inhibitory interneurons in cortical circuits.Front. Neural Circuits. 2016; 10: 76Crossref PubMed Scopus (104) Google Scholar, 13Lu J.T. Li C.Y. Zhao J.-P. Poo M.M. Zhang X.H. Spike-timing-dependent plasticity of neocortical excitatory synapses on inhibitory interneurons depends on target cell type.J. Neurosci. 2007; 27: 9711-9720Crossref PubMed Scopus (122) Google Scholar]. If excitatory synapses onto PV neurons are less likely to be potentiated when the neurons are targeted in the operant conditioning task, it can explain why the N− decrease strategy was employed with PV-INs. However, it has been shown that long-term potentiation can be induced in PV-INs with theta burst stimulation [14Sarihi A. Jiang B. Komaki A. Sohya K. Yanagawa Y. Tsumoto T. Metabotropic glutamate receptor type 5-dependent long-term potentiation of excitatory synapses on fast-spiking GABAergic neurons in mouse visual cortex.J. Neurosci. 2008; 28: 1224-1235Crossref PubMed Scopus (47) Google Scholar], and synaptic plasticity is sensitive to the neuromodulatory state of the circuit [15Huang S. Huganir R.L. Kirkwood A. Adrenergic gating of Hebbian spike-timing-dependent plasticity in cortical interneurons.J. Neurosci. 2013; 33: 13171-13178Crossref PubMed Scopus (38) Google Scholar] and behavioral context [16Donato F. Rompani S.B. Caroni P. Parvalbumin-expressing basket-cell network plasticity induced by experience regulates adult learning.Nature. 2013; 504: 272-276Crossref PubMed Scopus (433) Google Scholar, 17Kuhlman S.J. Olivas N.D. Tring E. Ikrar T. Xu X. Trachtenberg J.T. A disinhibitory microcircuit initiates critical-period plasticity in the visual cortex.Nature. 2013; 501: 543-546Crossref PubMed Scopus (268) Google Scholar]. Therefore, it remains unclear how the capacities for synaptic plasticity may differ across cell types in the intact brain during learning. Another potential mechanism for the improved performance in the current task is by modulating the activity of neurons presynaptic to the targeted neurons. Different subtypes of inhibitory neurons receive inputs from different populations of excitatory neurons. PV-INs receive dense inputs from nearby excitatory neurons [18Hofer S.B. Ko H. Pichler B. Vogelstein J. Ros H. Zeng H. Lein E. Lesica N.A. Mrsic-Flogel T.D. Differential connectivity and response dynamics of excitatory and inhibitory neurons in visual cortex.Nat. Neurosci. 2011; 14: 1045-1052Crossref PubMed Scopus (306) Google Scholar], which suggests that nearby PV-INs share similar excitatory inputs and thus a specific increase of excitatory inputs to one PV-IN may be difficult. On the other hand, SOM- and VIP-INs receive excitatory inputs from largely non-overlapping populations [19Karnani M.M. Jackson J. Ayzenshtat I. Tucciarone J. Manoocheri K. Snider W.G. Yuste R. Cooperative subnetworks of molecularly similar interneurons in mouse neocortex.Neuron. 2016; 90: 86-100Abstract Full Text Full Text PDF PubMed Scopus (112) Google Scholar]. Differences in the strategy for the operant conditioning task may originate from the differences in the characteristics of the neurons providing excitatory inputs to the target neurons. In addition, excitatory and inhibitory neurons form highly interconnected networks. In general, it is thought that VIP-INs inhibit SOM-INs; SOM-INs inhibit excitatory, PV-, and VIP-INs; PV-INs inhibit PV-INs and excitatory neurons [6Pfeffer C.K. Xue M. He M. Huang Z.J. Scanziani M. Inhibition of inhibition in visual cortex: the logic of connections between molecularly distinct interneurons.Nat. Neurosci. 2013; 16: 1068-1076Crossref PubMed Scopus (774) Google Scholar]; and excitatory neurons project to all four types. These connectivity patterns provide many possible pathways that could mediate the plasticity observed in the current study. For example, for SOM-INs, SOM → PV → excitatory → SOM provides a potential positive feedback loop, possibly underlying the increase of N+ activity in SOM-INs. Furthermore, the SOM → PV inhibition could underlie the decrease of N− activity in PV-INs. Future experiments are required to test these specific possibilities. Finally, the activity of the inhibitory neurons can be associated with movements and sensory stimulus in a subtype-specific manner. A study showed that monkeys perform a brain-machine interface task by exploring and exploiting neural patterns associated with natural movements [20Hwang E.J. Bailey P.M. Andersen R.A. Volitional control of neural activity relies on the natural motor repertoire.Curr. Biol. 2013; 23: 353-361Abstract Full Text Full Text PDF PubMed Scopus (59) Google Scholar]. Through our visual observations, we did not identify overt behavioral strategies during task performance, similar to a previous study [8Clancy K.B. Koralek A.C. Costa R.M. Feldman D.E. Carmena J.M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning.Nat. Neurosci. 2014; 17: 807-809Crossref PubMed Scopus (89) Google Scholar]. Nevertheless, examining the relationship between existing neural activity patterns and behavioral variables (“intrinsic neural manifold” and “intrinsic behavioral manifold”) [21Jazayeri M. Afraz A. Navigating the neural space in search of the neural code.Neuron. 2017; 93: 1003-1014Abstract Full Text Full Text PDF PubMed Scopus (124) Google Scholar] before training and how they change through training in each subtype will be of future interest. Furthermore, studies using neural feedback task with two-photon calcium imaging have reported that auditory [8Clancy K.B. Koralek A.C. Costa R.M. Feldman D.E. Carmena J.M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning.Nat. Neurosci. 2014; 17: 807-809Crossref PubMed Scopus (89) Google Scholar] or artificial sensory [22Prsa M. Galiñanes G.L. Huber D. Rapid integration of artificial sensory feedback during operant conditioning of motor cortex neurons.Neuron. 2017; 93: 929-939.e6Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar] feedback was necessary for the successful learning of the task, whereas similar learning only with rewards as a feedback has also been reported [23Hira R. Ohkubo F. Masamizu Y. Ohkura M. Nakai J. Okada T. Matsuzaki M. Reward-timing-dependent bidirectional modulation of cortical microcircuits during optical single-neuron operant conditioning.Nat. Commun. 2014; 5: 5551Crossref PubMed Scopus (17) Google Scholar]. Future experiments can be aimed at examining whether sensory feedback was necessary to learn to modulate inhibitory neurons and, if so, how the dependency is different among cell types. Lastly, we note that our contingency degradation experiments suggest that the behavioral performance was goal directed. However, the lack of auditory feedback in the degradation experiments leaves room for other interpretations, such as that the mice might have been in a completely different behavioral state without the feedback. It will be of future interest to investigate the relationships between targeted cell types and behavioral strategies, dependence on sensory feedback, and whether the behavioral performance is goal directed. Tabled 1REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesRabbit anti-PVAbcamCat#ab11427; RRID: AB_298032Chicken anti-GFPAves LabsCat#GFP-1020; RRID: AB_10000240Goat anti-chicken DyLight 488Thermo Fisher ScientificCat#SA5-10070; RRID: AB_2556650Donkey anti-rabbit Alexa 594Thermo Fisher ScientificCat#A-21207; RRID: AB_141637Experimental Models: Organisms/StrainsPV-CreThe Jackson LaboratoryRRID: IMSR_JAX:017320SOM-CreThe Jackson LaboratoryRRID: IMSR_JAX:013044VIP-CreThe Jackson LaboratoryRRID: IMSR_JAX:010908lsl-GCaMP6FThe Jackson LaboratoryRRID: IMSR_JAX:024105VIP-Cre::ZtTA::TITL-GCaMP6fThe Jackson LaboratoryRRID: IMSR_JAX:024107Software and AlgorithmsScanImage 4Vidrio TechnologiesRRID: SCR_014307MATLAB R2014aThe MathWorksRRID: SCR_001622 Open table in a new tab Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Takaki Komiyama ([email protected]). All procedures were in accordance with protocols approved by UCSD Institutional Animal Care and Use Committee and guidelines of the US National Institutes of Health. All animals before water restriction were group housed, and during water restriction they were singly housed or group housed when all littermates were under water restriction. They were housed in disposable plastic cages with standard bedding in a room on a reversed light cycle (12h/12h). Experiments were typically performed during the dark" @default.
• W2771907952 created "2017-12-22" @default.
• W2771907952 creator A5008557660 @default.
• W2771907952 creator A5054374543 @default.
• W2771907952 creator A5084052779 @default.
• W2771907952 date "2018-01-01" @default.
• W2771907952 modified "2023-10-16" @default.
• W2771907952 title "Brain-Computer Interface with Inhibitory Neurons Reveals Subtype-Specific Strategies" @default.
• W2771907952 cites W1966537851 @default.
• W2771907952 cites W1975032719 @default.
• W2771907952 cites W1988187967 @default.
• W2771907952 cites W1992910467 @default.
• W2771907952 cites W2017638232 @default.
• W2771907952 cites W2024332125 @default.
• W2771907952 cites W2025806474 @default.
• W2771907952 cites W2034709142 @default.
• W2771907952 cites W2038055149 @default.
• W2771907952 cites W2040574649 @default.
• W2771907952 cites W2044124598 @default.
• W2771907952 cites W2047378045 @default.
• W2771907952 cites W2047712100 @default.
• W2771907952 cites W2048272660 @default.
• W2771907952 cites W2061681507 @default.
• W2771907952 cites W2062159891 @default.
• W2771907952 cites W2064016110 @default.
• W2771907952 cites W2065582606 @default.
• W2771907952 cites W2089823661 @default.
• W2771907952 cites W2094125104 @default.
• W2771907952 cites W2101544546 @default.
• W2771907952 cites W2128675085 @default.
• W2771907952 cites W2294745036 @default.
• W2771907952 cites W2304994747 @default.
• W2771907952 cites W2526401721 @default.
• W2771907952 cites W2589192637 @default.
• W2771907952 cites W2594429360 @default.
• W2771907952 cites W808307776 @default.
• W2771907952 doi "https://doi.org/10.1016/j.cub.2017.11.035" @default.
• W2771907952 hasPubMedCentralId "https://www.ncbi.nlm.nih.gov/pmc/articles/5760288" @default.
• W2771907952 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/29249656" @default.
• W2771907952 hasPublicationYear "2018" @default.
• W2771907952 type Work @default.
• W2771907952 sameAs 2771907952 @default.
• W2771907952 citedByCount "23" @default.
• W2771907952 countsByYear W27719079522018 @default.
• W2771907952 countsByYear W27719079522019 @default.
• W2771907952 countsByYear W27719079522020 @default.
• W2771907952 countsByYear W27719079522021 @default.
• W2771907952 countsByYear W27719079522022 @default.
• W2771907952 countsByYear W27719079522023 @default.
• W2771907952 crossrefType "journal-article" @default.
• W2771907952 hasAuthorship W2771907952A5008557660 @default.
• W2771907952 hasAuthorship W2771907952A5054374543 @default.
• W2771907952 hasAuthorship W2771907952A5084052779 @default.
• W2771907952 hasBestOaLocation W27719079521 @default.
• W2771907952 hasConcept C113843644 @default.
• W2771907952 hasConcept C169760540 @default.
• W2771907952 hasConcept C17077164 @default.
• W2771907952 hasConcept C55352822 @default.
• W2771907952 hasConcept C55493867 @default.
• W2771907952 hasConcept C58226133 @default.
• W2771907952 hasConcept C86803240 @default.
• W2771907952 hasConceptScore W2771907952C113843644 @default.
• W2771907952 hasConceptScore W2771907952C169760540 @default.
• W2771907952 hasConceptScore W2771907952C17077164 @default.
• W2771907952 hasConceptScore W2771907952C55352822 @default.
• W2771907952 hasConceptScore W2771907952C55493867 @default.
• W2771907952 hasConceptScore W2771907952C58226133 @default.
• W2771907952 hasConceptScore W2771907952C86803240 @default.
• W2771907952 hasFunder F4320306079 @default.
• W2771907952 hasFunder F4320306148 @default.
• W2771907952 hasFunder F4320307384 @default.
• W2771907952 hasFunder F4320308324 @default.
• W2771907952 hasFunder F4320332161 @default.
• W2771907952 hasFunder F4320335353 @default.
• W2771907952 hasIssue "1" @default.
• W2771907952 hasLocation W27719079521 @default.
• W2771907952 hasLocation W27719079522 @default.
• W2771907952 hasLocation W27719079523 @default.
• W2771907952 hasOpenAccess W2771907952 @default.
• W2771907952 hasPrimaryLocation W27719079521 @default.
• W2771907952 hasRelatedWork W2019265405 @default.
• W2771907952 hasRelatedWork W2044962914 @default.
• W2771907952 hasRelatedWork W2147617498 @default.
• W2771907952 hasRelatedWork W2343911421 @default.
• W2771907952 hasRelatedWork W2626670817 @default.
• W2771907952 hasRelatedWork W2751847892 @default.
• W2771907952 hasRelatedWork W2951553096 @default.
• W2771907952 hasRelatedWork W3006582284 @default.
• W2771907952 hasRelatedWork W39906813 @default.
• W2771907952 hasRelatedWork W4212915683 @default.
• W2771907952 hasVolume "28" @default.
• W2771907952 isParatext "false" @default.
• W2771907952 isRetracted "false" @default.
• W2771907952 magId "2771907952" @default.
• W2771907952 workType "article" @default.