Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Distinct recurrent versus afferent dynamics in cortical visual processing

Nature Neuroscience volume 18, pages 1789–1797 (2015)Cite this article

Subjects

Abstract

How intracortical recurrent circuits in mammalian sensory cortex influence dynamics of sensory representation is not understood. Previous methods could not distinguish the relative contributions of recurrent circuits and thalamic afferents to cortical dynamics. We accomplish this by optogenetically manipulating thalamus and cortex. Over the initial 40 ms of visual stimulation, excitation from recurrent circuits in visual cortex progressively increased to exceed direct thalamocortical excitation. Even when recurrent excitation exceeded thalamic excitation, upon silencing thalamus, sensory-evoked activity in cortex decayed rapidly, with a time constant of 10 ms, which is similar to a neuron's integration time window. In awake mice, this cortical decay function predicted the time-locking of cortical activity to thalamic input at frequencies <15 Hz and attenuation of the cortical response to higher frequencies. Under anesthesia, depression at thalamocortical synapses disrupted the fidelity of sensory transmission. Thus, we determine dynamics intrinsic to cortical recurrent circuits that transform afferent input in time.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Time course of cortical recurrent excitation versus thalamic afferent excitation in visual cortex.
Figure 2: Time course of shut-off of visually evoked cortical activity during thalamic silencing.
Figure 3: Time course of shut-off of visually evoked activity in V1 and V2 of awake mice during thalamic silencing.
Figure 4: Relative contributions of thalamocortical synaptic depression and recurrent circuit dynamics to V1's response in anesthetized versus awake mice.
Figure 5: Cortical decay function is modulated by inhibition.

Similar content being viewed by others

References

  1. Ko, H. et al. Functional specificity of local synaptic connections in neocortical networks. Nature 473, 87–91 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Hubel, D.H. & Wiesel, T.N. Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. J. Physiol. (Lond.) 160, 106–154 (1962).

    Article  CAS  Google Scholar 

  3. Haider, B. et al. Synaptic and network mechanisms of sparse and reliable visual cortical activity during nonclassical receptive field stimulation. Neuron 65, 107–121 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Cossell, L. et al. Functional organization of excitatory synaptic strength in primary visual cortex. Nature (2015).

  5. Douglas, R.J., Koch, C., Mahowald, M., Martin, K.A. & Suarez, H.H. Recurrent excitation in neocortical circuits. Science 269, 981–985 (1995).

    CAS  PubMed  Google Scholar 

  6. Druckmann, S. & Chklovskii, D.B. Neuronal circuits underlying persistent representations despite time varying activity. Curr. Biol. 22, 2095–2103 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. van Vreeswijk, C. & Sompolinsky, H. Chaos in neuronal networks with balanced excitatory and inhibitory activity. Science 274, 1724–1726 (1996).

    CAS  PubMed  Google Scholar 

  8. Hawken, M.J., Shapley, R.M. & Grosof, D.H. Temporal-frequency selectivity in monkey visual cortex. Vis. Neurosci. 13, 477–492 (1996).

    Article  CAS  PubMed  Google Scholar 

  9. Movshon, J.A., Thompson, I.D. & Tolhurst, D.J. Spatial and temporal contrast sensitivity of neurones in areas 17 and 18 of the cat's visual cortex. J. Physiol. (Lond.) 283, 101–120 (1978).

    Article  CAS  Google Scholar 

  10. DeAngelis, G.C., Ohzawa, I. & Freeman, R.D. Spatiotemporal organization of simple-cell receptive fields in the cat's striate cortex. I. General characteristics and postnatal development. J. Neurophysiol. 69, 1091–1117 (1993).

    Article  CAS  PubMed  Google Scholar 

  11. Holub, R.A. & Morton-Gibson, M. Response of visual cortical neurons of the cat to moving sinusoidal gratings: response-contrast functions and spatiotemporal interactions. J. Neurophysiol. 46, 1244–1259 (1981).

    Article  CAS  PubMed  Google Scholar 

  12. Saul, A.B. & Humphrey, A.L. Temporal-frequency tuning of direction selectivity in cat visual cortex. Vis. Neurosci. 8, 365–372 (1992).

    Article  CAS  PubMed  Google Scholar 

  13. O'Keefe, L.P., Levitt, J.B., Kiper, D.C., Shapley, R.M. & Movshon, J.A. Functional organization of owl monkey lateral geniculate nucleus and visual cortex. J. Neurophysiol. 80, 594–609 (1998).

    Article  CAS  PubMed  Google Scholar 

  14. Foster, K.H., Gaska, J.P., Nagler, M. & Pollen, D.A. Spatial and temporal frequency selectivity of neurones in visual cortical areas V1 and V2 of the macaque monkey. J. Physiol. (Lond.) 365, 331–363 (1985).

    Article  CAS  Google Scholar 

  15. MacLean, J.N., Watson, B.O., Aaron, G.B. & Yuste, R. Internal dynamics determine the cortical response to thalamic stimulation. Neuron 48, 811–823 (2005).

    Article  CAS  PubMed  Google Scholar 

  16. Miller, J.E., Ayzenshtat, I., Carrillo-Reid, L. & Yuste, R. Visual stimuli recruit intrinsically generated cortical ensembles. Proc. Natl. Acad. Sci. USA 111, E4053–E4061 (2014).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Kloc, M. & Maffei, A. Target-specific properties of thalamocortical synapses onto layer 4 of mouse primary visual cortex. J. Neurosci. 34, 15455–15465 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  18. Castro-Alamancos, M.A. & Oldford, E. Cortical sensory suppression during arousal is due to the activity-dependent depression of thalamocortical synapses. J. Physiol. (Lond.) 541, 319–331 (2002).

    Article  CAS  Google Scholar 

  19. Borst, J.G. The low synaptic release probability in vivo. Trends Neurosci. 33, 259–266 (2010).

    Article  CAS  PubMed  Google Scholar 

  20. Swadlow, H.A., Gusev, A.G. & Bezdudnaya, T. Activation of a cortical column by a thalamocortical impulse. J. Neurosci. 22, 7766–7773 (2002).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Lien, A.D. & Scanziani, M. Tuned thalamic excitation is amplified by visual cortical circuits. Nat. Neurosci. 16, 1315–1323 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. da Costa, N.M. & Martin, K.A. The proportion of synapses formed by the axons of the lateral geniculate nucleus in layer 4 of area 17 of the cat. J. Comp. Neurol. 516, 264–276 (2009).

    Article  PubMed  Google Scholar 

  23. Li, Y.T., Ibrahim, L.A., Liu, B.H., Zhang, L.I. & Tao, H.W. Linear transformation of thalamocortical input by intracortical excitation. Nat. Neurosci. 16, 1324–1330 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Ferster, D., Chung, S. & Wheat, H. Orientation selectivity of thalamic input to simple cells of cat visual cortex. Nature 380, 249–252 (1996).

    Article  CAS  PubMed  Google Scholar 

  25. Poulet, J.F., Fernandez, L.M., Crochet, S. & Petersen, C.C. Thalamic control of cortical states. Nat. Neurosci. 15, 370–372 (2012).

    Article  CAS  PubMed  Google Scholar 

  26. Halassa, M.M. et al. Selective optical drive of thalamic reticular nucleus generates thalamic bursts and cortical spindles. Nat. Neurosci. 14, 1118–1120 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Constantinople, C.M. & Bruno, R.M. Effects and mechanisms of wakefulness on local cortical networks. Neuron 69, 1061–1068 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Timofeev, I., Grenier, F., Bazhenov, M., Sejnowski, T.J. & Steriade, M. Origin of slow cortical oscillations in deafferented cortical slabs. Cereb. Cortex 10, 1185–1199 (2000).

    Article  CAS  PubMed  Google Scholar 

  29. Nauhaus, I., Busse, L., Carandini, M. & Ringach, D.L. Stimulus contrast modulates functional connectivity in visual cortex. Nat. Neurosci. 12, 70–76 (2009).

    Article  CAS  PubMed  Google Scholar 

  30. Steriade, M., Timofeev, I. & Grenier, F. Natural waking and sleep states: a view from inside neocortical neurons. J. Neurophysiol. 85, 1969–1985 (2001).

    Article  CAS  PubMed  Google Scholar 

  31. Rager, G. & Singer, W. The response of cat visual cortex to flicker stimuli of variable frequency. Eur. J. Neurosci. 10, 1856–1877 (1998).

    Article  CAS  PubMed  Google Scholar 

  32. Murray, J.D. et al. A hierarchy of intrinsic timescales across primate cortex. Nat. Neurosci. 17, 1661–1663 (2014).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Niell, C.M. & Stryker, M.P. Modulation of visual responses by behavioral state in mouse visual cortex. Neuron 65, 472–479 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Niell, C.M. & Stryker, M.P. Highly selective receptive fields in mouse visual cortex. J. Neurosci. 28, 7520–7536 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Boudreau, C.E. & Ferster, D. Short-term depression in thalamocortical synapses of cat primary visual cortex. J. Neurosci. 25, 7179–7190 (2005).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Swadlow, H.A., Bezdudnaya, T. & Gusev, A.G. Spike timing and synaptic dynamics at the awake thalamocortical synapse. Prog. Brain Res. 149, 91–105 (2005).

    Article  PubMed  Google Scholar 

  37. Stark, E. et al. Inhibition-induced theta resonance in cortical circuits. Neuron 80, 1263–1276 (2013).

    Article  CAS  PubMed  Google Scholar 

  38. Han, X. et al. A high-light sensitivity optical neural silencer: development and application to optogenetic control of non-human primate cortex. Front. Syst. Neurosci. 5, 18 (2011).

    Article  PubMed  PubMed Central  Google Scholar 

  39. Atallah, B.V., Bruns, W., Carandini, M. & Scanziani, M. Parvalbumin-expressing interneurons linearly transform cortical responses to visual stimuli. Neuron 73, 159–170 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Fuster, J.M. Memory networks in the prefrontal cortex. Prog. Brain Res. 122, 309–316 (2000).

    Article  CAS  PubMed  Google Scholar 

  41. Gabernet, L., Jadhav, S.P., Feldman, D.E., Carandini, M. & Scanziani, M. Somatosensory integration controlled by dynamic thalamocortical feed-forward inhibition. Neuron 48, 315–327 (2005).

    Article  CAS  PubMed  Google Scholar 

  42. Fuster, J.M. & Alexander, G.E. Neuron activity related to short-term memory. Science 173, 652–654 (1971).

    Article  CAS  PubMed  Google Scholar 

  43. Camperi, M. & Wang, X.J. A model of visuospatial working memory in prefrontal cortex: recurrent network and cellular bistability. J. Comput. Neurosci. 5, 383–405 (1998).

    Article  CAS  PubMed  Google Scholar 

  44. Gold, J.I. & Shadlen, M.N. The neural basis of decision making. Annu. Rev. Neurosci. 30, 535–574 (2007).

    Article  CAS  PubMed  Google Scholar 

  45. Sherman, S.M. Thalamocortical interactions. Curr. Opin. Neurobiol. 22, 575–579 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  46. Bignall, K.E. & Rutledge, L.T. Origin of a photically evoked afterdischarge in cat visual cortex. J. Neurophysiol. 27, 1048–1062 (1964).

    Article  CAS  PubMed  Google Scholar 

  47. David, S.V., Vinje, W.E. & Gallant, J.L. Natural stimulus statistics alter the receptive field structure of v1 neurons. J. Neurosci. 24, 6991–7006 (2004).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. Muller, J.R., Metha, A.B., Krauskopf, J. & Lennie, P. Information conveyed by onset transients in responses of striate cortical neurons. J. Neurosci. 21, 6978–6990 (2001).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Nathan, J. et al. Scotopic and photopic visual thresholds and spatial and temporal discrimination evaluated by behavior of mice in a water maze. Photochem. Photobiol. 82, 1489–1494 (2006).

    CAS  PubMed  Google Scholar 

  50. Murray, J.D. et al. Linking microcircuit dysfunction to cognitive impairment: effects of disinhibition associated with schizophrenia in a cortical working memory model. Cereb. Cortex 24, 859–872 (2014).

    Article  PubMed  Google Scholar 

  51. Taniguchi, H. et al. A resource of Cre driver lines for genetic targeting of GABAergic neurons in cerebral cortex. Neuron 71, 995–1013 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Zhao, S. et al. Cell type-specific channelrhodopsin-2 transgenic mice for optogenetic dissection of neural circuitry function. Nat. Methods 8, 745–752 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Madisen, L. et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci. 13, 133–140 (2010).

    CAS  PubMed  Google Scholar 

  54. Hippenmeyer, S. et al. A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 3, e159 (2005).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  55. Boyden, E.S., Zhang, F., Bamberg, E., Nagel, G. & Deisseroth, K. Millisecond-timescale, genetically targeted optical control of neural activity. Nat. Neurosci. 8, 1263–1268 (2005).

    Article  CAS  PubMed  Google Scholar 

  56. Margrie, T.W., Brecht, M. & Sakmann, B. In vivo, low-resistance, whole-cell recordings from neurons in the anaesthetized and awake mammalian brain. Pflugers Arch. 444, 491–498 (2002).

    Article  CAS  PubMed  Google Scholar 

  57. Paxinos, G. & Franklin, K.B.J. The Mouse Brain in Stereotaxic Coordinates compact 2nd edn. (Elsevier Academic, 2004).

  58. Wang, Q., Gao, E. & Burkhalter, A. In vivo transcranial imaging of connections in mouse visual cortex. J. Neurosci. Methods 159, 268–276 (2007).

    Article  PubMed  Google Scholar 

  59. Fee, M.S., Mitra, P.P. & Kleinfeld, D. Automatic sorting of multiple unit neuronal signals in the presence of anisotropic and non-Gaussian variability. J. Neurosci. Methods 69, 175–188 (1996).

    Article  CAS  PubMed  Google Scholar 

  60. Adesnik, H., Bruns, W., Taniguchi, H., Huang, Z.J. & Scanziani, M. A neural circuit for spatial summation in visual cortex. Nature 490, 226–231 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  61. Goard, M. & Dan, Y. Basal forebrain activation enhances cortical coding of natural scenes. Nat. Neurosci. 12, 1444–1449 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  62. Mitra, P. & Bokil, H. Observed Brain Dynamics (Oxford Univ. Press, 2008).

  63. Abbott, L.F., Varela, J.A., Sen, K. & Nelson, S.B. Synaptic depression and cortical gain control. Science 275, 220–224 (1997).

    Article  CAS  PubMed  Google Scholar 

  64. Fuhrmann, G., Cowan, A., Segev, I., Tsodyks, M. & Stricker, C. Multiple mechanisms govern the dynamics of depression at neocortical synapses of young rats. J. Physiol. (Lond.) 557, 415–438 (2004).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank J. Evora for help with genotyping and mouse husbandry, as well as J. Isaacson, B. Bloodgood, C. Reinhold, S. Larson and members of the Scanziani and Isaacson labs for discussions about the project and comments on the manuscript. This project was supported by the National Science Foundation Graduate Research Program Fellowship, the Gatsby charitable foundation and the Howard Hughes Medical Institute.

Author information

Authors and Affiliations

  1. Neurosciences Graduate Program, University of California San Diego, La Jolla, California, USA

    Kimberly Reinhold, Anthony D Lien & Massimo Scanziani

  2. Center for Neural Circuits and Behavior, Neurobiology Section, University of California San Diego, La Jolla, California, USA

    Kimberly Reinhold, Anthony D Lien & Massimo Scanziani

  3. Department of Neuroscience, University of California San Diego, La Jolla, California, USA

    Kimberly Reinhold, Anthony D Lien & Massimo Scanziani

  4. Howard Hughes Medical Institute, University of California San Diego, La Jolla, California, USA

    Kimberly Reinhold, Anthony D Lien & Massimo Scanziani

Authors
  1. Kimberly Reinhold
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Anthony D Lien
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Massimo Scanziani
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

K.R. and M.S. designed the study. K.R. conducted all experiments and analyses, except the whole-cell recordings. A.D.L. performed and analyzed the whole-cell recordings. K.R. and M.S. wrote the paper.

Corresponding authors

Correspondence to Kimberly Reinhold or Massimo Scanziani.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Time course of thalamic and cortical responses to visual stimuli.

(a) Experimental configuration: whole-cell voltage-clamp of layer 4 (L4) neurons to record visually evoked excitatory post-synaptic currents (EPSCs) with or without optogenetic silencing of cortical excitatory recurrent circuits, as in Fig. 1b. Refers to (b). (b) Time course of synaptic excitation, comparing thalamic and cortical components. Left: Visual stimulus is static for 1.7 s. Right: Visual stimulus is static for 250 ms. Top: EPSCs from 6 example cells (arranged by thalamic fraction) in response to appearance of static grating (arrow), comparing control (black) versus cortical silencing with LED to isolate the thalamic EPSC (blue). Middle: Mean EPSC and s.e.m. across all cells presented with same-duration visual stimulus. Left average (n=8) includes both 1.7 s-long static gratings and 1.7 s-long luminance steps. Right average is response to 250 ms-long static gratings. Time until thalamic component is < 50% of total EPSC amplitude: (1.7 s-long stimuli) 45±13 ms, (250 ms-long stimuli) 42±8 ms (mean±s.e.m.). Bottom: Fraction of EPSC mediated by cortical excitation (cortical fraction, i.e., difference between black and blue, divided by black). Cortical fraction of the total integrated response: (1.7 s-long stimuli) 73±23%, mean±s.d., (250 ms-long stimuli) 62±23%.(c) Schematic of extracellular recordings in dLGN or V1 of anesthetized mice. (d) Time course of extracellular multi-unit activity in response to visual stimuli.Left column: Multi-unit activity in dLGN.Right column: Multi-unit activity in V1 (different animals than dLGN recordings at left).Top row: Average peri-stimulus time histogram (PSTH) of response to appearance of static grating (static for 3 s). V1 PSTH (right) is from Fig. 1c. Note that spike rate in V1 in response to the appearance of a static grating decreases more rapidly than the EPSC magnitude, possibly because intra-cortical synapses may depress more at the peak than later in the response 63,64 due to the higher spike frequency of cortical neurons at the peak response, because the multi-unit activity includes both excitatory and inhibitory neurons whereas the EPSC measures only excitation, because some component of the late-stage EPSC may originate outside of V1 (e.g., top-down input from other cortical areas), or for some combination of these reasons.

Second row: PSTH response to onset and offset of moving grating (moving for 3 s).Third row: PSTH response to appearance of static stimulus (both 3 s-long static gratings and 3 s-long luminance step). Bottom row: PSTH response to luminance step only (3 s duration, left) in dLGN or 10 ms-long luminance step (10 ms duration, occurring at 0 s) in V1 (right). Note that the response to even this very brief (10 ms-long) stimulus is long-lasting.

Supplementary Figure 2 Fitting the cortical decay function (CDF) in V1 of anesthetized mice.

For all parts, blue bar indicates LED illumination of TRN. Arrow indicates onset of visual stimulus.

(a) Left: Schematic of thalamic silencing by photo-activation of TRN. Right: (Top) TRN photo-activation prior to onset of visual stimulus prevents onset of a visual response in dLGN. (Bottom) TRN photo-activation suppresses ongoing visual response in LP. In these PSTHs, visual responses are to moving gratings (3 s), but silencing of the evoked response in dLGN was not different between responses to static or moving visual stimuli. At 300 ms into the response, as in Fig. 2c, the percentage of the evoked response suppressed in dLGN for static versus moving stimuli was: (static, n = 2) 93±57% (mean±s.e.m.), (moving, n = 12) 95±28%.

(b) Effect of thalamic silencing on spontaneous activity in V1 under anesthesia. Left to right: Effect of thalamic silencing on initiation of V1 Up states detected by LFP power ratio 63. (Left) Histograms of Up state multi-unit (MU) amplitude and duration when Up states begin during period of thalamic silencing; (Middle) probability of Up state initiation over 1 s window during thalamic silencing (blue) or control (black) (Up states detected in n = 14 mice); all p-values are from Wilcoxon rank-sum tests. (Right) Single-unit (SU) spontaneous activity from Up and Down states is summed within each recording session and normalized to 1. SU activity verifies that all spikes are from V1 neurons. Lines are mean, and error bars are s.d. of this spontaneous activity across mice. Spontaneous activity in V1 is not affected by silencing thalamus.

(c) Fitting V1 CDF. Top Row: To determine the exact time course of activity decay in V1 following thalamic silencing, i.e., the time course of the cortical decay function (CDF), we need to remove (deconvolve) the time course of shut-off in thalamus from the time course of shut-off in cortex. (Left) Deconvolution (dotted black) over 2-50 ms (using fits to dLGN = green and V1 = blue decays, see Methods). (Right) The deconvolution (dotted black) closely matches a 10 ms single exponential decay (blue) that begins at 3 ms after onset of the LED to silence the thalamus. Because the deconvolution introduces error in real, noisy data, we instead use a single exponential function, beginning at 3 ms after LED onset, to fit the CDF. Bottom Row: Fitting the CDF time course with two exponentials over longer timescales. (Left) Time course of average baseline-subtracted MU activity in V1 after silencing thalamus with LED. Dark blue: exponential fit component 1, a fast decay. Red: exponential fit component 2, a slow recovery component (red amplitude is 13% of dark blue). Gray dotted: sum of dark blue and red fits. (Right) As left panel but expanded timescale. Elsewhere in text we report a single exponential fit over initial 50-80 ms of the CDF, which is dominated by the fast component.

(d) Left: CDF across putative cortical layers (MU, n = 26 mice) and fits (dark blue). Right: Heat-map normalized across recording depths. 1 (red) is MU rate before LED onset; 0 (blue) is baseline rate. LED onset at 0 ms. Dotted lines show putative layer boundaries based on current source density (Methods). Black overlay: Mean±s.e.m. of time constant of single exponential fit to CDF. Note slightly slower shut-off in deeper layers.

(e) The CDF (mean norm. MU) is not affected by stimulus contrast, LED intensity or LED onset delay. Traces scaled and superimposed. Dark blue is 11 ms fit from Fig. 2f. Left: Contrast 1: light blue. Contrast 0.25: dark blue (n = 5 mice). Inset: Zoomed out MU evoked response in example mouse (thalamic silencing not complete) to low- (gray/dark blue) or high-contrast (light blue/black) moving grating with (light blue/dark blue) or without (black/gray) thalamic silencing; y axis: 200 Hz, x axis: 800 ms. Middle: As left but varying LED intensity to achieve 30-50%, 50-70%, 70-90%, or > 90% suppression of V1 evoked response (n = 2 mice). Inset: (left inset) V1 CDFs un-normalized, (right inset) different levels of dLGN silencing during LED intensities 0.5, 3.6 and 6.5 mW; y axis: 150 Hz, x axis: 1.5 s, PSTHs smoothed with 250 ms window. Right: As left but for LED onset delays 200 ms, 300 ms or 900 ms relative to stimulus onset (n = 3 mice). Inset: CDFs separated by onset delay, left to right: 200, 300, 900 ms. Supplementary Table 1 for more conditions.

(f) Example SU during visual stimulation at LED onset. Light blue shading: LED on. Gray line: mean baseline of each unit. Unit examples arranged by putative layer (see labels at left). All traces normalized to peak.

(g) Left: Mean norm. shut-off of regular-spiking (RS) units across cortical layers (red is fit). Inset shows mean and s.d. of waveforms for RS and fast-spiking (FS) units (Methods for separation of these unit types). Right: As at (left) but for fast-spiking (FS) units (blue is fit).(h) SU activity at moment of silencing thalamus. Each point is a unit. Top-Left: Spontaneous Up states in V1 are unaffected. Top-Right: Visually evoked activity is suppressed (3 s-long moving visual stimulus). Bottom-Left: Analysis of whether each single unit in V1 returns to its own spontaneous baseline after thalamus is silenced during a sensory response. X axis is each unit’s spontaneous rate (no visual stimulus, blank screen) during the LED, and Y axis is that unit’s evoked response (moving grating present) during the LED. No systematic change from pre-visual stimulus spontaneous activity when thalamus is off during visual stimulus.

(i) Comparison of V1 CDF with V1 activity at offset of moving grating (from Supplementary Fig. 1d) on expanded timescale. Black and white angled bars: moving grating stimulus (3 s duration), followed by transition to gray screen. Blue curve: fit to shut-off of visually evoked activity following optogenetic silencing of thalamus (scaled to match level of activity before shut-off onset). Note much faster shut-off of visually evoked activity after silencing thalamus, with respect to shut-off in V1 after just turning off the visual stimulus.

Supplementary Figure 3 dLGN inhibition correlates with shut-off of visually evoked response in V1.

(a) Expression of fluorescent reporter TdTomato fused to Flexed-ChR2 in Gad2-Cre mice in two sets of coronal sections from two mice (first mouse: left two sections and left scatter, second mouse: right two sections and right scatter). Coverage is the fraction of pixels expressing TdTomato within region-of-interest (ROI) outlining given brain area. Scatter plots compare reporter coverage of TRN region with reporter coverage of dLGN. For brain structures with > 1 mm anterior-posterior extent, coverage is average of coverage across three coronal sections spanning extent of structure. Coverage of TRN’s caudo-dorsal sector (putative visual sector) but not TRN’s somatosensory sector correlates with coverage of dLGN. ρ is correlation coefficient; p is p-value of correlation.

(b) Coverage, same as in (a), of different brain areas plotted against fractional suppression of V1 evoked response (frac. supp.) following photo-activation of TRN in anesthetized animals. Each point is one mouse. ρ is correlation coefficient; p is p-value of correlation. Coverage of TRN (visual sector) and dLGN are the only expression patterns necessary to observe shut-off of V1 evoked response (i.e., linear fit passes near the origin and has a positive slope). Note that expression of ChR2 in TRN axons to LP is not necessary to observe shut-off of V1 evoked response; even when ChR2 expression in axons to LP is weak, we may observe a strong shut-off of V1 evoked response. Abbreviations: V1 = primary visual cortex, A1 = primary auditory cortex, TRN = thalamic reticular nucleus, dLGN = dorsal lateral geniculate nucleus, vLGN = ventral lateral geniculate nucleus, IGL = intergeniculate leaflet, LP = lateral posterior nucleus, SC = superior colliculus, Auditory thal. = medial geniculate nucleus, S1 = primary somatosensory cortex, Somato. thal. = ventral posterior medial/lateral nuclei.

Supplementary Figure 4 Fitting the CDF in awake mice.

(a) Awake extracellular recordings in V1 or V2 and optogenetic configuration to silence thalamus by photo-activating TRN.

(b) Single-unit (SU) activity at moment of silencing thalamus. Each point is a unit. Dotted lines are unity. Points below unity line are suppressed. Left: Visually evoked activity (pre-stimulus baseline-subtracted) in V1 of awake mice is suppressed. Points below zero (solid horizontal gray line) are suppressed below baseline. Middle: Spontaneous activity in V1 of awake mice is also suppressed by silencing thalamus. Not baseline-subtracted. Right: Visually evoked activity (baseline-subtracted) in V2 of awake mice is suppressed. Points below zero (solid horizontal gray line) indicate suppression below baseline.

(c) V1 recording. Cortical decay function (CDF) is independent of animal’s behavioral state (left; superimposed MU shut-off for running and non-running trials; n = 4 mice), stimulus diameter (middle; as MU; 5, 7, 15 degrees or full-field; n = 3 mice), and single unit’s stimulus preference (i.e., grating orientation; mean SU during stimulation with preferred versus non-preferred orientation of moving grating; n = 127 SU). Dark blue fit is 11 ms, as in Fig. 2f. See Supplementary Table 1 for more conditions.

(d) V1 recording.

Left: Single exponential fit (blue) to cortical decay function (MU, black) across putative cortical layers (n = 11 awake mice). Right: Histogram of time constants of single exponential fits to CDF (MU) across conditions. This shows that CDF distribution is approximately normal. For parametric statistical tests of the CDF across conditions reported in paper, CDF distribution was assumed to be normal and variances were assumed to be equal across groups, but this was not formally tested. Count is mice. See Supplementary Table 1 for further break-down.

Supplementary Figure 5 CDF predicts amplitude of V1 response to high- and low-contrast flicker stimuli while recording sequentially or simultaneously in dLGN and V1.

(a) Top: Anesthetized. Frequency response under anesthesia of regular-spiking (RS) units in Layer 4 (black; n = 30 units; mean and s.e.m.). dLGN (green) and predicted V1 (pink) frequency response from left panel of Fig. 4c. Middle: Awake. Frequency response (n = 30 units) of RS units in Layer 4 (black; n = 30 units; mean and s.e.m.). dLGN (green) and predicted V1 (pink) frequency response from right panel of Fig. 4c. Bottom: Awake. Frequency response of RS throughout all layers (black; n = 270 units; mean and s.e.m.). dLGN (green) and predicted V1 (pink) frequency response from right panel of Fig. 4c. Note that, consistent with V1 response curves in Fig. 4, we observe theta-band amplification (4-11 Hz) 37 in V1 response (see Supplementary Fig. 7).

(b) Attenuation predicted by CDF over full response amplitude distribution. Frequency response as percentiles of single-unit (SU) amplitudes (peak-norm., see Methods for alignment, normalization and scaling). Left: Percentiles of dLGN frequency response (green) vs. same percentiles of V1 RS SU frequency response (black). Note that V1 amp. is below dLGN amp. at high frequencies for all percentiles of response amplitude distributions. Right: Black as left. Pink is percentiles of prediction for V1 by convolving dLGN frequency response with CDF. Shaded areas are differences between curves. At all percentiles, prediction for V1 after filtering by CDF is much better match to real V1 response than dLGN response without filtering by CDF.

(c) Deconvolution of dLGN PSTH response to flicker stimulus from V1 PSTH response to same stimulus (across all temporal frequencies, Methods) in anesthetized (Top) and awake (Bottom) mice. Gray line: mean deconvolution across all thalamus-V1 pairs with s.e.m. (gray-blue error). Black: single exponential fit and time constant value. Note much slower dynamics under anesthesia as compared to awake.

(d) Frequency response of dLGN and V1 to low contrast (20%) frequency chirps (vis. stimuli are frequency-modulated sweeps, shown below in (f)). Simultaneous dLGN and V1 recordings. Green = median frequency response of dLGN units and 45th to 55th percentile error bars, as in Fig. 4c. Black = mean, s.e.m. of regular-spiking (RS) cells across layers. Pink = prediction for V1, i.e., green filtered by CDF. Zero is noise (for alignment and scaling, see Methods).

(e) Example single RS unit response to 14 Hz flicker. Top: raster, Bottom: PSTH. Inset: unit waveform. Yellow: close-ups.

(f) Simultaneous extracellular recordings in dLGN and V1. Visual stimulus is logarithmic frequency modulation of full-field luminance (chirp). Unless specified, vertical scale bars are 100 Hz. Top traces: Superimposed MU PSTH from dLGN (green) and V1 (black) of example mouse in response to luminance modulation shown below (Vis. stim.). Middle traces: V1 response prediction (pink; by convolution of dLGN response with CDF) superimposed on actual V1 response (black, from top). Note that pink predicts extent of amplitude filtering at higher frequencies. Bottom traces: Pink, black: Expanded timescale of segment A (above). Green, black: Averaging MU across mice (left) or including only ON-responsive units (right) improves coherence between dLGN and V1 traces. Responses are to chirp sections marked B and C. (g) Coherence between dLGN and V1 recorded simultaneously. Black: mean and s.e.m. of coherence (between mean MU dLGN vs. V1 within each mouse) across mice. Red: Coherence between averaged-across-mice MU responses in dLGN and V1. Purple: Coherence between ON-responsive dLGN and V1 units in example mouse.

Supplementary Figure 6 Waking up from anesthesia increases baseline spiking in the thalamus.

(a) Average thalamic single-unit (SU) spontaneous firing rate over time as mice wake up from isoflurane anesthesia (5 curves are 5 example recordings; anesth. average SU rate = 0.8 Hz, awake average SU rate = 4.2 Hz). Light green: recording site in LP; black: 4 sites in dLGN.

(b) Left: Average spontaneous SU firing rate under anesthesia and awake. Right: Normalized to final rate in awake. Colors as in (a).

(c) Average spontaneous SU firing rate under anesthesia (isoflurane) versus SU average rate in awake state. Dotted gray line is unity. Points above dotted line have higher firing rates in the awake state. Light green dots are units in LP. Black dots are units in dLGN.

Supplementary Figure 7 Theta-band amplification by the CDF.

The slow, small recovery component of the CDF (dip below baseline, see Supplementary Fig. 2c) predicts theta-band amplification in V1’s frequency response.(a) CDF in time domain. Top: Pink: CDF of RS units. Curve is zeroed at pre-stimulus baseline activity (arrow). Note both fast shut-off, fit by 12 ms decaying single exponential, and second small, slow recovery component. Dark blue, dotted blue, solid gray, dotted gray: single exponential decays with time constants as indicated; 1 and 100 ms time constants shown for comparison. Bottom: Magnitude of slow recovery component as a function of where the baseline of CDF is set. The higher the baseline (dotted gray line), the more pronounced the slow recovery component of CDF (from red to brown). In this study, we choose the baseline of the CDF (under anesthesia) to be the mean pre-stimulus baseline, because thalamic silencing does not affect spontaneous activity under anesthesia. However, in each experiment, the CDF relaxes to a baseline that depends on the strength of thalamic silencing; thus, the baseline of the CDF may be offset with respect to the pre-stimulus baseline. Curves are offset on ordinate for clarity. Inset: green and pink from Fig. 4c (right). Arrow shows CDF amplification of stimulus frequencies in theta band.

(b) CDF represented in the frequency domain. Top: Pink: amplitude frequency spectrum of pink curve in (a). This pink filter is used to predict V1’s response. Colors as in (a) but spectra are in frequency domain (all scaled). Bottom: Frequency-domain forms of CDFs in bottom panel of (a) (colors are matched). Increasing the magnitude of the slow recovery component of CDF increases theta-band amplification (i.e., the 4-11 Hz component of the frequency-domain form of the CDF).

Supplementary Figure 8 Both fundamental (F1) and first harmonic (F2) components of V1 response are consistent with filtering by CDF.

This figure rules out an alternate model where responses at the fundamental (F1) frequency in dLGN are converted to first harmonics (F2) in V1 without filtering by the cortical decay function (CDF). We find that both F1 and F2 components in V1 are attenuated, consistent with filtering by the CDF. A transformation from F1 in dLGN to F2 in V1 should instead increase F2 in V1.

(a) Experimental set-up to measure frequency response in awake mice recording in either dLGN or V1.(b) Left: Heatmap is median power of dLGN single-unit response to flicker stimulus in awake mice as a function of stimulus frequency and response frequency. F1 response is the diagonal where response frequency matches stimulus frequency. F2 is the diagonal where response frequency is twice the stimulus frequency. Right: Same as left but for V1 regular-spiking (RS) units in awake.(c) Amplitude of F1 and F2 response components from heatmaps in (a). Includes all units, including units that are well-driven, as well as units that are not well-driven, by the stimulus, thus the low spike rate. Left: for dLGN relay units. Right: for V1 RS units. Consistent with data from cats 31, V1’s F1 response dips between 10 and 20 Hz, while the F2 response dips around 8 Hz.(d) Simple models of dLGN-to-V1 connectivity affecting spatial receptive field structures. Model 1: F1 response in dLGN drives F1 response in V1. No consideration of F2. This is the model presented in Fig. 4. Model 2A: F1 response in dLGN drives some F1 in V1 and some F2 in V1. F1-to-F2 transformation between dLGN and V1 is consistent with certain models 2 of the visual system. Also, F2 response in dLGN drives some F2 in V1. In Model 2A, no amplitude filtering by the CDF. Model 2B: Same as Model 2A, but Model 2B does include amplitude filtering by the CDF. See Methods for equations associated with models.

(e) Fraction of F1 response in dLGN that frequency-doubles in cortex to produce an F2 response in V1, derived from Model 2A (top) or 2B (bottom). See Methods for associated equations and derivation. Y axis is the fraction of response that is frequency-doubled as a function of stimulus frequency, according to best fit models. Above 30 Hz enters noise (so not shown). Dotted gray is average across stimulus frequencies.(f) Different predictions for V1 F1 and F2 response components by Models 1, 2A and 2B. Left: Prediction by Model 1, as in Fig. 4. Middle: Best prediction for both F1 and F2 components of V1 response according to Model 2A. Prediction does not match actual V1 F1 response. Right: Best prediction for both F1 and F2 components of V1 response according to Model 2B. By including filtering by the CDF in this model, this model can better account for the average amplitude of both the F1 and F2 response components in V1 RS cells.

Supplementary Figure 9 Slowing of V1 CDF correlates with widespread ArchT expression in inhibitory interneurons of visual cortex.

(a) All mice tested, subdivided according to ArchT expression profile in V1. Top Row: Expression of ArchT-GFP in visual cortex. Overlay of one coronal section from each mouse in Group. Grp. 1: widespread ArchT (n = 5 mice; these are 5 mice in Fig. 5), Grp. 2: deep layers (n = 8 mice), Grp. 3: superficial layers (n = 2 mice), Grp. 4: radially restricted (n = 2 mice). Row 2: Increase in gain of V1 multi-unit (MU) activity by ArchT photo-stimulation. Black: control, Red: ArchT photo-stimulation. Row 3: V1 MU activity with thalamic silencing. Blue: control, Purple: ArchT photo-stimulation. Arrow is onset of moving grating. Blue horizontal bar (blue LED) is photo-activation of TRN (thalamic silencing). Red horizontal bar (amber LED) is ArchT photo-stimulation. Error bars are s.d. across PSTH. Row 4: Time course of V1 MU (baseline-subtracted) after silencing thalamus. Blue: control, Purple: ArchT photo-stimulation. Bottom: Quantification of shut-off time course (stats. are two-sided paired t-tests; data distributions were assumed to be normal and variances were assumed to be equal across groups, but this was not formally tested – see figure for all data points).

(b) Coverage (fraction of pixels expressing ArchT within visual cortex) vs. change in time to half-max of V1 shut-off as a result of ArchT photo-stimulation. Grp. 1 includes the 5 mice with greatest ArchT coverage of visual cortex. R is correlation coefficient; p is p-value of correlation. Note strong correlation between expression of ArchT and slowing of CDF.(c) Same as (b), but metric is time constant of single exponential fit to V1 shut-off. R is correlation coefficient; p is p-value of correlation. Note strong correlation.(d) Quantification across all mice tested. Significant slowing of V1 shut-off with ArchT photo-stimulation across all mice (p = 0.00018) and even across mice when excluding the 5 with the greatest ArchT coverage (i.e., excluding 5 with the strongest effect on CDF, p = 0.02). Two-sided paired t-tests comparing interleaved conditions. Data distributions were assumed to be normal and variances were assumed to be equal across groups, but this was not formally tested.

Supplementary Figure 10 Reducing cortical inhibition to slow the CDF slows onset of V1 response.

(a) Schematic of dual optogenetic manipulation, as in Fig. 5.(b) Two example single units at moment of thalamic silencing with (purple) or without (blue) V1 disinhibition.(c) No further effect of silencing thalamus during cortical disinhibition on spontaneous activity in V1. Dark purple curve is CDF during cortical disinhibition from Fig. 5d. Light purple: Same time window as dark purple, i.e., activity in V1 at moment of silencing thalamus and during cortical disinhibition, but here without visual stimulus (spontaneous activity).

(d) Suppressing cortical inhibitory interneurons does not differentially increase cortical sensitivity to low, relative to high, levels of thalamic activity.

Fig. 5 suggests that suppressing the activity of inhibitory interneurons in cortex leads to prolonged dynamics (CDF) within cortical recurrent circuits. To show that the site of the prolonged CDF is cortical recurrent circuits, we need to rule out a change in cortical sensitivity to thalamic input as an alternate explanation for the prolonged CDF. If thalamic silencing were incomplete and if suppressing cortical inhibitory interneurons were to increase the gain of V1’s response differentially to low levels of thalamic input, we would expect a larger response in V1 when levels of thalamic input are low, that is, after onset of the blue LED to suppress activity in thalamus. This change in the transfer function between thalamus and cortex could explain the prolonged CDF observed during cortical disinhibition only if the gain of the V1 response is higher at low levels of thalamic input than it is at high levels of thalamic input. If the gain of the V1 response is increased by cortical disinhibition equally for high and low levels of thalamic input, then we expect the activity in V1 to be scaled up by a constant factor. Scaling the CDF by a constant factor does not change its time course. Thus, a simple change in gain of V1’s response to thalamic input cannot explain the observed change in V1 dynamics. Here we show a control experiment indicating that cortical disinhibition does not make V1 more sensitive to low, relative to high, levels of thalamic input. Thus, the transfer function between thalamus and cortex as a function of the level of activity is only scaled by a constant factor. Therefore the source of prolonged cortical dynamics (the prolonged CDF) is likely mechanisms of cortex independent of thalamic input.

First Two Columns: Multi-unit (MU) PSTH responses recorded in V1 for different interleaved LED conditions. Both the blue LED and the amber LED turn on before the visual stimulus in this experiment. Black: Control (no V1 disinhibition). Red-Black: Amber LED to suppress inhibitory interneurons in cortex (V1 disinh.). Blue-Black: Low thalamic activity is achieved by low-intensity blue LED illumination of TRN over first 500 ms of visually evoked response. Red-Blue: Both amber and blue LEDs (to partially suppress thalamic activity during V1 disinhibition). Left: No suppression of visually evoked response in thalamus; thus, response to high-contrast visual stimulus (moving grating onset) is high thalamic activity. Two-headed white arrow indicates increase in visually evoked activity at steady-state (gray shaded area) upon V1 disinhibition. Middle: Lowering thalamic activity through partial suppression of visually evoked response in thalamus by blue LED. Two-headed black arrow indicates increase in visually evoked activity at steady-state (gray shaded area) upon V1 disinhibition. Right: V1 disinhibition does not produce a greater gain increase at low, relative to high, levels of thalamic activity. Amplitude of cortical activity during low or high thalamic activity. Circle colors same as colors in plots to left. Gain, computed during steady-state (gray shaded area) is (black arrow) Red-Black over Black or (white arrow) Red-Blue over Blue-Black. Circles show mean response amplitude at steady-state (gray shaded areas), normalized to the response in control (black circle), across 3 mice.

(e) Slowing the CDF slows V1’s onset response. (i) Effect of control (blue) or slowed (purple) CDF on time course of response in V1 (MU) to visual stimulus onset. Interleaved LED conditions. Same mice as in Fig. 5d. Note lagged onset response following V1 disinhibition. (ii) Purple curve is from left. Lighter purple: Prediction of lagged V1 response to stimulus onset given the measured change in the CDF during V1 disinhibition (Methods). (iii) Quantification of average lag (n = 5 mice) over the initial 50 ms of the onset response (black) and of predicted lag (light purple). Positive values indicate delayed onset response. Note good match between experimental (black) and predicted lag (light purple). Left axis y-axis shows histogram of experimental (black) and predicted (light purple) lags. (iv) Slowing of single-unit (SU) onsets upon slowing the CDF, computed in a manner analogous to Fig. 5f scatter plot (the plot shows the ratio of SU firing rate during initial 50 ms of the onset response, called time window c, divided by firing rate during the subsequent 50 ms, which was 50-100 ms after response onset, called time window a). (v) Cartoon of onset response showing how higher values of this ratio (c over a) for control (blue) indicate faster rise in control conditions vs. V1 disinh. (purple). P-value (two-sided paired t-test; data distribution was assumed to be normal and variances were assumed to be equal across groups, but this was not formally tested – see figure for all data points) for SU slowing is p = 0.04.

(f) Increasing gain of V1 response by varying vis. stim. contrast (Left, gray = high contrast, black = low contrast) or behavioral state (Right, gray = running, black = non-running) leads to faster, not slower, rise of onset response.

Supplementary Figure 11 Rate-limiting afferent or recurrent dynamic control of responses in visual cortex.

Sites of rate-limiting processes are colored in red. Under anesthesia (left), depression at thalamic afferent synapses is rate-limiting and strongly low-pass filters input from dLGN. In awake mice (right), both low- and high-frequency input from dLGN pass equally well through thalamic afferent synapses. In the awake state, the dynamics of intra-cortical recurrent circuits become rate-limiting.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–11 and Supplementary Table 1 (PDF 2266 kb)

Supplementary Methods Checklist (PDF 592 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Reinhold, K., Lien, A. & Scanziani, M. Distinct recurrent versus afferent dynamics in cortical visual processing. Nat Neurosci 18, 1789–1797 (2015). https://doi.org/10.1038/nn.4153

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nn.4153

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing