Overlearning hyperstabilizes a skill by rapidly making neurochemical processing inhibitory-dominant

  • A Corrigendum to this article was published on 01 October 2017

Abstract

Overlearning refers to the continued training of a skill after performance improvement has plateaued. Whether overlearning is beneficial is a question in our daily lives that has never been clearly answered. Here we report a new important role: overlearning in humans abruptly changes neurochemical processing, to hyperstabilize and protect trained perceptual learning from subsequent new learning. Usually, learning immediately after training is so unstable that it can be disrupted by subsequent new learning until after passive stabilization occurs hours later. However, overlearning so rapidly and strongly stabilizes the learning state that it not only becomes resilient against, but also disrupts, subsequent new learning. Such hyperstabilization is associated with an abrupt shift from glutamate-dominant excitatory to GABA-dominant inhibitory processing in early visual areas. Hyperstabilization contrasts with passive and slower stabilization, which is associated with a mere reduction of excitatory dominance to baseline levels. Using hyperstabilization may lead to efficient learning paradigms.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Figure 1: Procedures and results of Experiment 1.
Figure 2: Procedures and results of Experiment 2.
Figure 3: Procedures and results of Experiment 3.

Change history

  • 18 September 2017

    In the version of this article initially published, NIH grant R01EY019466 was missing from grants to T.W. in the Acknowledgments. The error has been corrected in the HTML and PDF versions of the article.

References

  1. 1

    Ebbinghaus, H. Über das Gedächtnis. Untersuchungen zur experimentellen Psychologie (Duncker & Humblot, Leipzig, 1885).

  2. 2

    Pashler, H., Rohrer, D., Cepeda, N.J. & Carpenter, S.K. Enhancing learning and retarding forgetting: choices and consequences. Psychon. Bull. Rev. 14, 187–193 (2007).

    PubMed  Google Scholar 

  3. 3

    Sasaki, Y., Nanez, J.E. & Watanabe, T. Advances in visual perceptual learning and plasticity. Nat. Rev. Neurosci. 11, 53–60 (2010).

    CAS  PubMed  Google Scholar 

  4. 4

    Brashers-Krug, T., Shadmehr, R. & Bizzi, E. Consolidation in human motor memory. Nature 382, 252–255 (1996).

    CAS  PubMed  Google Scholar 

  5. 5

    Breton, J. & Robertson, E.M. Flipping the switch: mechanisms that regulate memory consolidation. Trends Cogn. Sci. 18, 629–634 (2014).

    PubMed  Google Scholar 

  6. 6

    Dewar, M.T., Cowan, N. & Sala, S.D. Forgetting due to retroactive interference: a fusion of Müller and Pilzecker's (1900) early insights into everyday forgetting and recent research on anterograde amnesia. Cortex 43, 616–634 (2007).

    PubMed  PubMed Central  Google Scholar 

  7. 7

    Mosha, N. & Robertson, E.M. Unstable memories create a high-level representation that enables learning transfer. Curr. Biol. 26, 100–105 (2016).

    CAS  PubMed  Google Scholar 

  8. 8

    Robertson, E.M., Pascual-Leone, A. & Miall, R.C. Current concepts in procedural consolidation. Nat. Rev. Neurosci. 5, 576–582 (2004).

    CAS  PubMed  Google Scholar 

  9. 9

    Seitz, A.R. et al. Task-specific disruption of perceptual learning. Proc. Natl. Acad. Sci. USA 102, 14895–14900 (2005).

    CAS  PubMed  Google Scholar 

  10. 10

    Petroff, O.A. GABA and glutamate in the human brain. Neuroscientist 8, 562–573 (2002).

    CAS  PubMed  Google Scholar 

  11. 11

    Crist, R.E., Li, W. & Gilbert, C.D. Learning to see: experience and attention in primary visual cortex. Nat. Neurosci. 4, 519–525 (2001).

    CAS  PubMed  Google Scholar 

  12. 12

    Furmanski, C.S., Schluppeck, D. & Engel, S.A. Learning strengthens the response of primary visual cortex to simple patterns. Curr. Biol. 14, 573–578 (2004).

    CAS  PubMed  Google Scholar 

  13. 13

    Gilbert, C.D., Li, W. & Piech, V. Perceptual learning and adult cortical plasticity. J. Physiol. (Lond.) 587, 2743–2751 (2009).

    CAS  Google Scholar 

  14. 14

    Li, W., Piëch, V. & Gilbert, C.D. Perceptual learning and top-down influences in primary visual cortex. Nat. Neurosci. 7, 651–657 (2004).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15

    Li, W., Piëch, V. & Gilbert, C.D. Learning to link visual contours. Neuron 57, 442–451 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16

    Schoups, A., Vogels, R., Qian, N. & Orban, G. Practising orientation identification improves orientation coding in V1 neurons. Nature 412, 549–553 (2001).

    CAS  PubMed  Google Scholar 

  17. 17

    Shibata, K., Watanabe, T., Sasaki, Y. & Kawato, M. Perceptual learning incepted by decoded fMRI neurofeedback without stimulus presentation. Science 334, 1413–1415 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  18. 18

    Yotsumoto, Y., Watanabe, T. & Sasaki, Y. Different dynamics of performance and brain activation in the time course of perceptual learning. Neuron 57, 827–833 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  19. 19

    Keppel, G. & Underwood, B.J. Proactive inhibition in short-term retention of single items. J. Verbal Learn. Verbal Behav. 1, 153–161 (1962).

    Google Scholar 

  20. 20

    Krakauer, J.W. Motor learning and consolidation: the case of visuomotor rotation. Adv. Exp. Med. Biol. 629, 405–421 (2009).

    PubMed  PubMed Central  Google Scholar 

  21. 21

    Yotsumoto, Y., Chang, L.H., Watanabe, T. & Sasaki, Y. Interference and feature specificity in visual perceptual learning. Vision Res. 49, 2611–2623 (2009).

    PubMed  PubMed Central  Google Scholar 

  22. 22

    Barron, H.C. et al. Unmasking latent inhibitory connections in human cortex to reveal dormant cortical memories. Neuron 90, 191–203 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  23. 23

    Cohen Kadosh, K., Krause, B., King, A.J., Near, J. & Cohen Kadosh, R. Linking GABA and glutamate levels to cognitive skill acquisition during development. Hum. Brain Mapp. 36, 4334–4345 (2015).

    PubMed  PubMed Central  Google Scholar 

  24. 24

    Hensch, T.K. Critical period plasticity in local cortical circuits. Nat. Rev. Neurosci. 6, 877–888 (2005).

    CAS  PubMed  Google Scholar 

  25. 25

    Kim, S., Stephenson, M.C., Morris, P.G. & Jackson, S.R. tDCS-induced alterations in GABA concentration within primary motor cortex predict motor learning and motor memory: a 7 T magnetic resonance spectroscopy study. Neuroimage 99, 237–243 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  26. 26

    Letzkus, J.J., Wolff, S.B. & Lüthi, A. Disinhibition, a circuit mechanism for associative learning and memory. Neuron 88, 264–276 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  27. 27

    Lunghi, C., Emir, U.E., Morrone, M.C. & Bridge, H. Short-term monocular deprivation alters GABA in the adult human visual cortex. Curr. Biol. 25, 1496–1501 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  28. 28

    Morishita, H. & Hensch, T.K. Critical period revisited: impact on vision. Curr. Opin. Neurobiol. 18, 101–107 (2008).

    CAS  Google Scholar 

  29. 29

    Stagg, C.J., Bachtiar, V. & Johansen-Berg, H. The role of GABA in human motor learning. Curr. Biol. 21, 480–484 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  30. 30

    Vallentin, D., Kosche, G., Lipkind, D. & Long, M.A. Neural circuits. Inhibition protects acquired song segments during vocal learning in zebra finches. Science 351, 267–271 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  31. 31

    Banai, K., Ortiz, J.A., Oppenheimer, J.D. & Wright, B.A. Learning two things at once: differential constraints on the acquisition and consolidation of perceptual learning. Neuroscience 165, 436–444 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  32. 32

    Tartaglia, E.M., Aberg, K.C. & Herzog, M.H. Perceptual learning and roving: Stimulus types and overlapping neural populations. Vision Res. 49, 1420–1427 (2009).

    PubMed  Google Scholar 

  33. 33

    Martin, D.L. & Rimvall, K. Regulation of gamma-aminobutyric acid synthesis in the brain. J. Neurochem. 60, 395–407 (1993).

    CAS  PubMed  Google Scholar 

  34. 34

    Stagg, C.J. Magnetic resonance spectroscopy as a tool to study the role of GABA in motor-cortical plasticity. Neuroimage 86, 19–27 (2014).

    CAS  PubMed  Google Scholar 

  35. 35

    Stagg, C.J. & Rothman, D.L. Magnetic Resonance Spectroscopy (Academic Press, 2013).

  36. 36

    Belelli, D. et al. Extrasynaptic GABAA receptors: form, pharmacology, and function. J. Neurosci. 29, 12757–12763 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  37. 37

    Myers, J.F.M., Evans, C.J., Kalk, N.J., Edden, R.A.E. & Lingford-Hughes, A.R. Measurement of GABA using J-difference edited 1H-MRS following modulation of synaptic GABA concentration with tiagabine. Synapse 68, 355–362 (2014).

    CAS  PubMed  Google Scholar 

  38. 38

    Beppu, K. et al. Optogenetic countering of glial acidosis suppresses glial glutamate release and ischemic brain damage. Neuron 81, 314–320 (2014).

    CAS  PubMed  Google Scholar 

  39. 39

    Okubo, Y. et al. Imaging extrasynaptic glutamate dynamics in the brain. Proc. Natl. Acad. Sci. USA 107, 6526–6531 (2010).

    CAS  PubMed  Google Scholar 

  40. 40

    Hauptmann, B., Reinhart, E., Brandt, S.A. & Karni, A. The predictive value of the leveling off of within session performance for procedural memory consolidation. Brain Res. Cogn. Brain Res. 24, 181–189 (2005).

    PubMed  Google Scholar 

  41. 41

    Censor, N., Karni, A. & Sagi, D. A link between perceptual learning, adaptation and sleep. Vision Res. 46, 4071–4074 (2006).

    PubMed  Google Scholar 

  42. 42

    Ofen, N., Moran, A. & Sagi, D. Effects of trial repetition in texture discrimination. Vision Res. 47, 1094–1102 (2007).

    PubMed  Google Scholar 

  43. 43

    Sagi, D. Perceptual learning in vision research. Vision Res. 51, 1552–1566 (2011).

    Google Scholar 

  44. 44

    Regan, D. & Beverley, K.I. Postadaptation orientation discrimination. J. Opt. Soc. Am. 2, 147–155 (1985).

    CAS  Google Scholar 

  45. 45

    Shibata, K., Sagi, D. & Watanabe, T. Two-stage model in perceptual learning: toward a unified theory. Ann. NY Acad. Sci. 1316, 18–28 (2014).

    PubMed  Google Scholar 

  46. 46

    Bang, J.W., Khalilzadeh, O., Hämäläinen, M., Watanabe, T. & Sasaki, Y. Location specific sleep spindle activity in the early visual areas and perceptual learning. Vision Res. 99, 162–171 (2014).

    PubMed  Google Scholar 

  47. 47

    Mascetti, L. et al. The impact of visual perceptual learning on sleep and local slow-wave initiation. J. Neurosci. 33, 3323–3331 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  48. 48

    Yotsumoto, Y. et al. Location-specific cortical activation changes during sleep after training for perceptual learning. Curr. Biol. 19, 1278–1282 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  49. 49

    Hasselmo, M.E. & McClelland, J.L. Neural models of memory. Curr. Opin. Neurobiol. 9, 184–188 (1999).

    CAS  PubMed  Google Scholar 

  50. 50

    Shadmehr, R. & Holcomb, H.H. Neural correlates of motor memory consolidation. Science 277, 821–825 (1997).

    CAS  PubMed  PubMed Central  Google Scholar 

  51. 51

    Heba, S. et al. Local GABA concentration predicts perceptual improvements after repetitive sensory stimulation in humans. Cereb. Cortex 26, 1295–1301 (2016).

    PubMed  Google Scholar 

  52. 52

    Seitz, A.R., Kim, D. & Watanabe, T. Rewards evoke learning of unconsciously processed visual stimuli in adult humans. Neuron 61, 700–707 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  53. 53

    Xiao, L.Q. et al. Complete transfer of perceptual learning across retinal locations enabled by double training. Curr. Biol. 18, 1922–1926 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  54. 54

    Wandell, B.A., Dumoulin, S.O. & Brewer, A.A. Visual field maps in human cortex. Neuron 56, 366–383 (2007).

    CAS  PubMed  Google Scholar 

  55. 55

    Hu, Y., Chen, X., Gu, H. & Yang, Y. Resting-state glutamate and GABA concentrations predict task-induced deactivation in the default mode network. J. Neurosci. 33, 18566–18573 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  56. 56

    Mescher, M., Merkle, H., Kirsch, J., Garwood, M. & Gruetter, R. Simultaneous in vivo spectral editing and water suppression. NMR Biomed. 11, 266–272 (1998).

    CAS  PubMed  Google Scholar 

  57. 57

    Rothman, D.L., Behar, K.L., Hetherington, H.P. & Shulman, R.G. Homonuclear 1H double-resonance difference spectroscopy of the rat brain in vivo. Proc. Natl. Acad. Sci. USA 81, 6330–6334 (1984).

    CAS  PubMed  Google Scholar 

  58. 58

    Hancu, I. Optimized glutamate detection at 3T. J. Magn. Reson. Imaging 30, 1155–1162 (2009).

    PubMed  PubMed Central  Google Scholar 

  59. 59

    Mullins, P.G., Chen, H., Xu, J., Caprihan, A. & Gasparovic, C. Comparative reliability of proton spectroscopy techniques designed to improve detection of J-coupled metabolites. Magn. Reson. Med. 60, 964–969 (2008).

    CAS  PubMed  Google Scholar 

  60. 60

    Tkác, I., Starcuk, Z., Choi, I.Y. & Gruetter, R. In vivo 1H NMR spectroscopy of rat brain at 1 ms echo time. Magn. Reson. Med. 41, 649–656 (1999).

    PubMed  Google Scholar 

  61. 61

    Harris, A.D. et al. Impact of frequency drift on gamma-aminobutyric acid-edited MR spectroscopy. Magn. Reson. Med. 72, 941–948 (2014).

    CAS  PubMed  Google Scholar 

  62. 62

    Robertson, C.E., Ratai, E.M. & Kanwisher, N. Reduced GABAergic action in the autistic brain. Curr. Biol. 26, 80–85 (2016).

    CAS  PubMed  Google Scholar 

  63. 63

    Provencher, S.W. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn. Reson. Med. 30, 672–679 (1993).

    CAS  PubMed  Google Scholar 

  64. 64

    Provencher, S.W. Automatic quantitation of localized in vivo 1H spectra with LCModel. NMR Biomed. 14, 260–264 (2001).

    CAS  PubMed  Google Scholar 

  65. 65

    Kreis, R. Issues of spectral quality in clinical 1H-magnetic resonance spectroscopy and a gallery of artifacts. NMR Biomed. 17, 361–381 (2004).

    CAS  PubMed  Google Scholar 

  66. 66

    Stagg, C.J. et al. Polarity-sensitive modulation of cortical neurotransmitters by transcranial stimulation. J. Neurosci. 29, 5202–5206 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  67. 67

    Brainard, D.H. The psychophysics toolbox. Spat. Vis. 10, 433–436 (1997).

    CAS  Google Scholar 

Download references

Acknowledgements

We thank A. Berard, J. Dobres, M. Nassar, D. Rahnev and E. Robertson for their important comments on early drafts. This work was supported by NIH R01 EY015980 and R01EY019466 (to T.W.), NSF BCS 1539717 (to Y.S.) and JSPS KAKENHI Grant Number 16H06857 (to K.S.). L.-H.C. was supported by MOST (104-2410-H-010-001-MY2, 105-2420-H-010-002-MY2), NYMU Aging and Health Research Center and Yen Tjing Ling Medical Foundation.

Author information

Affiliations

Authors

Contributions

K.S., Y.S., E.G.W., M.G.M., M.T. and T.W. designed the experiments. K.S., J.W.B., M.G.M., M.T. and L.-H.C. conducted the experiments. K.S., E.G.W., M.T. and M.G.M. analyzed data. K.S., Y.S., J.W.B., E.G.W. and T.W. wrote the manuscript.

Corresponding author

Correspondence to Takeo Watanabe.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Schematic illustration of our results.

Overlearning hyper-stabilizes learning states by rapidly changing the E/I ratio in the early visual areas from more excitatory to more inhibitory than the pre-training baseline, so that learning of a task will not be disrupted by subsequent training on a new task. Several hours after training, irrespective of whether hyper-stabilization occurred or not, the E/I ratio returns to the baseline, leading to typical stabilization which allows for both the existing and new learning without disrupting each other.

Supplementary Figure 2 Procedures and results of the preliminary experiment.

(a) Schematic of a trial of the orientation detection task. Subjects were asked to indicate which of the 2 stimulus intervals contained an oriented structure (2-interval-forced-choice task). (b) Results. The purpose of the preliminary experiment was to estimate the amount of training that induces the saturation of performance improvements of the orientation detection task (Supplementary Fig. 2a) so that the effects of overlearning on learning stability would be tested in the main experiments. The experiment consisted of 3 stages conducted over 2 consecutive days: pre-test, training, and post-test stages (see Preliminary experiment in Online Methods for details). Subjects were asked to perform the orientation detection task (Supplementary Fig. 2a). Task difficulty was controlled by adjusting the signal-to-noise (S/N) ratio of the stimuli using a standard 2-down 1-up staircase method. S/N ratio thresholds were measured for each block of trials. Five different groups of subjects (N=12 for each group) were trained on the task for one orientation (trained orientation) for 4, 7, 8, 9, and 16 blocks, respectively. During each of the pre- and post-test stages, we measured S/N ratio thresholds for 3 orientations (10, 70, and 130 degrees): one was the trained orientation, while the remaining 2 served as untrained orientations that were rotated ±60 degrees away from the trained orientation. Performance improvement for each orientation after training was calculated as percent reduction in the S/N ratio threshold in the post-test stage relative to the pre-test stage (see Pre- and post-test stages in Online Methods for the definition of performance improvement). The graph shows the mean (±s.e.m.) performance improvement for the trained orientation as a function of the number of training blocks. We applied segmented regression analysis on the mean performance improvements. The root mean square error for the 2 linearly interpolated lines connected at one breaking point was calculated when the breaking point was the 7th, 8th, or 9th block of training. The root mean square error was the smallest when the breaking point was at the 8th block. The 2 interpolated lines with the 8th block as the breaking point are shown as black lines in the graph. These results suggest that the performance improvement plateaued around the 8-block training. That is, the 8-block training group experienced no overlearning, while the 16-block training group experienced overlearning.

Supplementary Figure 3 Procedures and results of Control Experiments 1 and 2.

(a) Procedure of Control Experiment 1. (b) Procedure of Control Experiment 2 in which the time interval between the first- and second-training phases was set to 50 min. (c) Mean (±s.e.m.) performance improvements (N=12) for the first-trained, second-trained, and untrained orientations in Control Experiment 1. A significant performance improvement was found only for the first-trained orientation (one-sample t-test, t11=3.615, P=0.012 after Bonferroni correction for 3 comparisons), but not for the second-trained (t11=0.705, P=0.496 without Bonferroni correction) or untrained (t11=0.578, P=0.575 without Bonferroni correction) orientation. (d) Mean (±s.e.m.) performance improvements (N=12) for the first-trained, second-trained, and untrained orientations in Control Experiment 2. A significant performance improvement was found only for the second-trained orientation (one-sample t-test, t11=3.953, P=0.007 after Bonferroni correction for 3 comparisons), but not for the first-trained (t11=1.008, P=0.335 without Bonferroni correction) or untrained (t11=0.706, P=0.495 without Bonferroni correction) orientation. * P<0.05, ** P<0.01 after Bonferroni correction.

Supplementary Figure 4 Example spectra from the voxel located at the early visual areas.

The measured spectra are shown in the top row. The second row shows the spectra fitted with the LC-Model (see MRS analysis in Online Methods for details). The bottom row shows the residual remaining after the fitting. The remaining rows show individual fits for all metabolites that can be detected by a given acquisition. Macromolecular and lipid signals were used to produce the baseline correction. (a) An example spectrum from the GABA scan. (b) An example spectrum from the glutamate scan. NAA, N-acetylaspartate; NAAG, N-acetylaspartylglutamate; GSH, glutathione; Glu, Glutamate; GABA, gamma-aminobutyric acid; Asp, aspartate; GPC, glycerophosphocholine; Ins, myo-inositol; Lac, lactate; Glc, glucose; Gln, Glutamine, PCh, phosphocholine; Cr, creatine; CrCH2, creatine methylene group; PCr, phosphocreatine; Scyllo, scyllo-inositol; Tau, taurine.

Supplementary Figure 5 Raw spectra for the overlearning group (n = 12) in Experiment 3.

The value on each plot shows the full-width-at-half-maximum line-width for NAA in Hz. (a) The raw spectra obtained in the glutamate scan for each of 12 subjects and each of pre-MRS, post-MRS 1, and post-MRS 2 stages. (b) The raw difference spectra obtained in the GABA scan.

Supplementary Figure 6 Performance improvements in Experiment 3.

(a) Mean (±s.e.m.) performance improvements in the no-overlearning group for the trained and 2 untrained orientations (orientation rotated -60 degrees from the trained, orientation rotated +60 degrees from the trained). A significant performance improvement was found only for the trained orientation (one-sample t-test, t11=4.905, P=0.003 after Bonferroni correction for 6 comparisons), and not for the untrained orientations (t11<1.292, P>0.223 without Bonferroni correction across both tests). (b) Mean (±s.e.m.) performance improvements in the overlearning group. A significant performance improvement was found only for the trained orientation (t11=4.483, P=0.006 after Bonferroni correction for 6 comparisons), and not for the untrained orientations (t11<0.840, P>0.419 without Bonferroni correction across both tests). In addition, the performance improvements for the trained orientation did not significantly differ between the 2 groups (two-sample t-test, t22=0.510, P=0.615). ** P<0.01 after Bonferroni correction.

Supplementary Figure 7 Results of additional analyses of MRS data in Experiment 3.

(a) Mean (±s.e.m.) E/I ratio changes calculated using an alternative method. While some MRS studies have used glutamate as a representative excitatory neurotransmitter (Jocham et al., Nat Neurosci, 2012; Terhune et al., J Neurosci, 2014), others have used a combined signal from glutamate and glutamine (Glx) (Ref. 55; Duncan et al., PLoS One, 2013). We tested whether the use of Glx, instead of glutamate alone, affects the observed E/I ratio changes. We observed the same statistical tendency, as originally found in Experiment 3 (Fig. 3d). Results of a two-way mixed-model ANOVA on E/I ratio change with factors of time (30 min after vs. 3.5 hours after training) and group (no-overlearning vs. overlearning groups) showed a significant main effect of group (F1,22=17.886, P<10-3) and a significant interaction between time and group (F1,22=5.281, P=0.031). No significant main effect of time was observed (F1,22=0.009, P=0.924). In the no-overlearning group (red), we found a significant quadratic trend in the time-course of the E/I ratio change (F1,11=5.270, P=0.042). The E/I ratio change was significantly greater than zero 30 min after training (one-sample t-test, t11=3.543, P=0.018 after Bonferroni correction for 4 comparisons), but not significantly different from zero 3.5 hours after training (t11=0.947, P=0.364 without Bonferroni correction). In the overlearning group (cyan), we also found a significant quadratic trend in the time-course of the E/I ratio change (F1,11=7.278, P=0.021). The E/I ratio change was significantly less than zero 30 min after training (one-sample t-test, t11=3.041, P=0.045 after Bonferroni correction for 4 comparisons), but not significantly different from zero 3.5 hours after training (t11=1.251, P=0.237 without Bonferroni correction). * P<0.05 after Bonferroni correction. (b) Mean (±s.e.m.) percent changes in the creatine concentrations for the no-overlearning (red) and overlearning (cyan) groups. We did not find any evidence that group or time significantly affected creatine concentrations. Results of a two-way mixed-model ANOVA on change in the creatine concentrations with factors being time (30 min after vs. 3.5 hours after training) and group (no-overlearning vs. overlearning groups) showed no significant main effect of time (F1,22=0.413, P=0.527), group (F1,22=0.248, P=0.623), or interaction between the 2 factors (F1,22=1.067, P=0.313). In addition, no significant change in the creatine concentration from the baseline was found 30 min (one-sample t-test, t11<0.783, P>0.450) or 3.5 hours (t11<0.751, P>0.468) after training for either group.

Supplementary Figure 8 Procedures and results of Control Experiment 3.

(a) Procedures. During the training stage, 2 orientations were presented on alternating blocks during 16 blocks of training (red and cyan box). (b) Mean (±s.e.m.) performance improvements (N=7) for trained orientation 1, trained orientation 2, and untrained orientation. No significant performance improvement was found for any orientation (one-sample t-test, t6<0.296, P>0.777). (c) Mean (±s.e.m.) E/I ratio changes. The E/I ratio change was not significantly different from zero 30 min after training (one-sample t-test, t6=1.074, P=0.324) or 3.5 hours after training (t6=0.731, P=0.492).

Supplementary Figure 9 Procedures and results of Experiment 4.

(a) Procedures. During the first-training stage (red box), subjects were trained on the orientation detection task with one orientation (first-trained orientation) for 16 blocks. During the second-training stage (cyan box), subjects were trained on the task with another orientation (second-trained orientation) for 8 blocks. (b) Mean (±s.e.m.) performance improvements (N=12) for the first-trained, second-trained, and untrained orientations. A significant performance improvement was found only for the first-trained orientation (one-sample t-test, t11=4.968, P=0.001 after Bonferroni correction for 3 comparisons), but not for the second-trained (t11=1.132, P=0.282 without Bonferroni correction) or untrained (t11=0.369, P=0.722 without Bonferroni correction) orientation. These results indicate that anterograde interference occurred, as originally found in Experiment 1. ** P<0.01 after Bonferroni correction. (c) Mean (±s.e.m.) E/I ratio changes. As in Experiment 3, the E/I ratio was significantly decreased 30 min after the 16-block training in the first-training stage (one-sample t-test, t11=3.179, P=0.009). ** P<0.01. (d) The scatter plot of the E/I ratio change 30 min after the first-training stage against the performance improvement for the second-trained orientation. The black line indicates the ordinary least-square regression. A significant correlation was found between the E/I ratio change 30 min after the first-training stage and the performance improvement for the second-trained orientation across subjects (r=0.643, P=0.024). There was no significant outlier in the plot (Grubbs’ test, P>0.05).

Supplementary Figure 10 Changes in glutamate and GABA in Experiment 3.

(a) Mean (±s.e.m.) changes in the concentrations of glutamate (magenta) and GABA (blue) for the overlearning group (N=12). A two-way ANOVA with repeated measures on concentration change with factors of time (30 min vs. 3.5 hours after training) and metabolite (glutamate vs. GABA) showed a significant main effect for metabolite (F1,11=9.934, P=0.009), but not for time (F1,11=0.326, P=0.580) nor interaction between the 2 factors (F1,11=1.827, P=0.204). For glutamate, no significant quadratic trend in the time-course of the glutamate changes was found (F1,11=0.002, P=0.967). Subsequent tests showed no significant change in the glutamate concentration 30 min after (one-sample t-test, t11=0.143, P=0.889) or 3.5 hours after (t11=0.172, P=0.867) training. For GABA, we also found no significant quadratic trend in the time-course of the GABA changes (F1,11=4.189, P=0.065). Subsequent tests showed no significant change in the GABA concentration 30 min after (one-sample t-test, t11=2.637, P=0.092 after Bonferroni correction for 4 comparisons) or 3.5 hours after training (t11=1.314, P=0.216). (b) Mean (±s.e.m.) changes in the concentrations of glutamate and GABA for the no-overlearning group (N=12). A two-way ANOVA with repeated measures on concentration change with factors of time (30 min vs. 3.5 hours after training) and metabolite (glutamate vs. GABA) showed a significant main effect for metabolite (F1,11=7.175, P=0.022), but not for time (F1,11=0.426, P=0.528) nor interaction between the 2 factors (F1,11=1.471, P=0.251). For glutamate, no significant quadratic trend in the time-course of the glutamate changes was found (F1,11=0.820, P=0.385). Subsequent tests showed no significant change in the glutamate concentration 30 min after (one-sample t-test, t11=1.973, P=0.074) or 3.5 hours after (t11=1.403, P=0.188) training. For GABA, we also found no significant quadratic trend in the time-course of the GABA changes (F1,11=0.206, P=0.659). Subsequent tests showed no significant change in the GABA concentration 30 min after (one-sample t-test, t11=0.506, P=0.623) or 3.5 hours after (t11=1.204, P=0.254) training.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–10 and Supplementary Tables 1 and 2 (PDF 1593 kb)

Supplementary Methods Checklist (PDF 676 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Shibata, K., Sasaki, Y., Bang, J. et al. Overlearning hyperstabilizes a skill by rapidly making neurochemical processing inhibitory-dominant. Nat Neurosci 20, 470–475 (2017). https://doi.org/10.1038/nn.4490

Download citation

Further reading