doi: 10.1038/nn.4601. Epub 2017 Jul 10.
Neural reactivations during sleep determine network credit assignment
Tanuj Gulati 1 2 3, Ling Guo 1 2 3, Dhakshin S Ramanathan 1 3 4 5, Anitha Bodepudi 1 2, Karunesh Ganguly 1 2 3
Affiliations
- PMID:28692062
- PMCID: PMC5808917
- DOI: 10.1038/nn.4601
Item in Clipboard
Neural reactivations during sleep determine network credit assignment
Tanuj Gulati et al. Nat Neurosci.2017 Sep.
Display options
Format
Display options
Format
Abstract
A fundamental goal of motor learning is to establish the neural patterns that produce a desired behavioral outcome. It remains unclear how and when the nervous system solves this 'credit assignment' problem. Using neuroprosthetic learning, in which we could control the causal relationship between neurons and behavior, we found that sleep-dependent processing was required for credit assignment and the establishment of task-related functional connectivity reflecting the casual neuron-behavior relationship. Notably, we observed a strong link between the microstructure of sleep reactivations and credit assignment, with downscaling of non-causal activity. Decoupling of spiking to slow oscillations using optogenetic methods eliminated rescaling. Thus, our results suggest that coordinated firing during sleep is essential for establishing sparse activation patterns that reflect the causal neuron-behavior relationship.
Conflict of interest statement
None
Figures
a, The practice sessions were separated by a block of sleep. Rats learned direct neural control of a feeding tube (θ = angular position). Successful trials required movement fromP1 toP2 within 15 s.b, A typical trial structure is depicted. c, Comparison of trial times. A significant reduction in completion time was found betweenBMI1Late toBMI2Early (n = 10 sessions; pairedt test,t9 = 7.62, *P < 10−4).d, At the top are the waveforms and inter-spike interval histograms of the neurons analyzed below (color-coded). Plot below shows the trend in the modulation depth ratio (MDratio) during BMI performance for three neurons before and after sleep. Another neuron whose waveform is not shown is depicted in green. Below are the peri–event histograms fromBMI1Late andBMI2Early trials, respectively for theTRD andTRI neurons (in same color convention). Thick line represents mean; shaded area is the jackknife error. Below the PETHs are representative spike rasters from multiple trials. Red dot indicates task completion time for each trial.e, Average modulation depth change (MD∆) betweenBMI1 andBMI2 (mean in solid line ± s.e.m. in box; unpairedt tests;BMI1 andBMI2Early t121 = 6.79, **P < 10−9;BMI1 andBMI2Late t121 = 6.31, ***P < 10−8;BMI1 andBMI2 t121 = 6.96, **P < 10−9).
a, Example plot of SSC as a function of frequency during sleep prior to (Sleeppre) and after (Sleeppost forTRD – TRD; red forTRD – TRI pairs) skill acquisition. The lighter band is the jackknife error. The box highlights the 0.3 – 4 Hz band.b, Relationship between SSC change before and after learning, and change in task-related modulation after sleep,MDΔ(BMI1Late toBMI2Early), spearman correlation,r(123) = 0.51,P < 10−8.c, Average modulation depth during reactivations (MDreactivation, i.e. ratio of peak to tails) ofTRD neurons fromSleeppre toSleeppost.d,MDreactivation ofTRI neurons fromSleeppre toSleeppost.e, Average modulation depth duringSleeppre toSleeppost reactivations forTRD and TRI neurons (mean in solid line ± s.e.m. in box, one-way ANOVA,F3,242 = 34.28,P < 10−17; significantpost hoc t tests, *P < 0.05).
a, Neural firing centered to task start and task end/reward for the same session for regular BMI training (i.e.BMIfixed-reward). The lighter band is the jackknife error.b, Schematic of “variable-reward” BMI training.b, Schematic of variable-reward BMI trials.c, Average Fano factor ofTRD andTRI neurons for the four sets of conditions, namely task-start (successful and unsuccessful trials are separately parsed) and task-end/reward frame inBMIfixed-reward, and task end inBMIvariable-reward (mean in solid line ± s.e.m. in box, task start and task end inBMIfixed-reward one-way ANOVA,F5,350 = 41.20,P < 10−32; task end inBMIfixed-reward andBMIvariable-reward one-way ANOVA,F3,166 = 83.86,P < 10−32, significantpost hoc t tests, *P < 0.05).
a, Pairwise correlation of neural firing forTRD – TRD andTRD – TRI pairs around task start and task end inBMIfixed-reward andBMIvariable-reward paradigms (mean in solid line ± s.e.m. in box; one-way ANOVA,F7,304 = 8.36,P < 10−8; significantpost hoc t tests, *P < 0.05).b, Relationship of individual neural pairwise (i.e. at task end) and reactivation during sleep inBMIfixed-reward sessions (linear regressionR2 = 0.54,P < 10−21; neural pairs are in same convention as Fig 4a).c, Relationship of individual neural pairwise correlations at task end and reactivation during sleep inBMIvariable-reward sessions (linear regressionR2 = 0.07,P > 0.05; neural pairs are in same convention as Fig 4a).
a, Fluorescence image of a coronal brain section showing neurons expressing Jaws (green) in M1. Scale bar is 500 μm.b, UP state triggered LED inhibition of aTRD cell inSleeppost as compared to the activity of same cell inSleeppre without stimulation. Rasters are shown along with raw traces of the local-field potential (LFPs) based on threshold crossing of the LFP. Dark line is the mean LFP. Bottom-most row shows histogram of firing activity.c, Top: Average modulation depth (MD) of aTRD cell in a representativeOPTOUP experiment. Bottom: Average modulation depth (MD) ofTRD cells around slow-oscillations inOPTOUP,OPTODOWN, andOPTOOFF experiments (mean in solid line ± s.e.m. in box, one-way ANOVA,F2,41 = 425.75,P < 10−27; significantpost hoc t tests, *P < 0.05).d, Examples of the raw and filtered (0.3–4 Hz) traces and the stimulation period for respectiveOPTOUP andOPTODOWN experiments.e, Power spectrum of LFP fromSleeppre andSleeppost in anOPTOUP experiments. The lighter band is the jackknife error.f, Power spectral changes (in 0.3 – 4 Hz) forOPTOUP,OPTODOWN, andOPTOOFF experiments (one-way ANOVA,F2,27 = 0.13,P = 0.87).
a, Learning curves from twoBMI sessions in the same rat with and without optogenetic inhibition during sleep (i.e.OPTOUP andOPTOOFF sessions, respectively).b, Performance changes fromBMI1Late toBMI2Early in each of the three respective conditions (OPTOUP sessions pairedt testt10 = -5.52, *P < 10−3;OPTODOWN sessions pairedt testt7 = 5.12, *P < 10−3;OPTOOFF sessions pairedt testt7 = 7.73, **P < 10−4).
a, Rescaling ofTRD andTRI neurons measured through modulation depth change (MD∆) fromBMI1 andBMI2 inOPTOUP,OPTODOWN, andOPTOOFF experiments (mean in solid line ± s.e.m. in box;OPTOUP sessions unpairedt testt110 = −0.47,P = 0.64;OPTODOWN sessions unpairedt testt106 = 3.67, *P < 10−3;OPTOOFF sessions pairedt testt73 = 5.52, **P < 10−6).b, Example plot of SFC as a function of frequency inSleeppre andSleeppost inOPTOUP andOPTODOWN experiment for twoTRD neurons. The lighter band is the jackknife error.c, Averaged SFC changes fromSleeppre toSleeppost for TRD neuronsin OPTOUP,OPTODOWN, andOPTOOFF groups (mean in solid line ± s.e.m. in box, one-way ANOVA,F2,41 = 44.83,P < 10−10; significantpost hoc t tests, ***P < 0.05).d, Averaged SFC changes forTRD cells versus averaged rescaling ofTRI cells fromBMI1 toBMI2 in OPTOUP,OPTODOWN, andOPTOOFF groups (linear regressionR2 = 0.66,P < 10−6).
References
MeSH terms
Grants and funding
LinkOut - more resources
Full Text Sources
Other Literature Sources