A neural circuit mechanism for regulating vocal variability during song learning in zebra finches

  1. Jonathan Garst-Orozco
  2. Baktash Babadi
  3. Bence P Ölveczky  Is a corresponding author
  1. Harvard University, United States
10 figures and 2 tables

Figures

Probing the neural mechanisms underlying the regulation of motor variability in songbirds.

(A) A strong coupling between performance improvements and variability reduction is a hallmark of most forms of motor skill learning. (B) Spectral derivatives of songs from a single zebra finch at …

https://doi.org/10.7554/eLife.03697.003
Experimental approach for probing inputs to RA projection neurons from LMAN and HVC during different stages of song development.

(A) Bright-field image of an acute brain slice encompassing RA and incoming fibers from HVC (parasaggital slice). The fibers are stimulated and evoked currents from RA projection neurons recorded in …

https://doi.org/10.7554/eLife.03697.004
Inputs from HVC to RA neurons change throughout song development.

(A) Schematic highlighting the inputs being probed (red box). (B) Distributions of SF currents for the three age groups tested. Red line represents the log-normal fit to the data (‘Materials and …

https://doi.org/10.7554/eLife.03697.005
Inputs from LMAN to RA neurons remain largely unchanged throughout song development.

(A) Schematic highlighting the inputs being probed (green box). (B) Currents evoked in response to stimulating the LMAN fiber tract with an intensity resulting in either failures (grey) or EPSCs of …

https://doi.org/10.7554/eLife.03697.007
Intrinsic properties of RA projection neurons do not change significantly with song development.

(A) To test intrinsic excitability of RA neurons as a function of age we injected current into RA cells and measured the membrane voltage in current clamp. (B) Membrane voltage in an RA projection …

https://doi.org/10.7554/eLife.03697.009
A simple model relates strengthening and pruning of connections in a motor control network (here: HVC-RA) to reduced motor variability.

(A) Top: the early phase of song learning is characterized by high vocal variability (left panel). This is associated with relatively larger number of weaker inputs from HVC to RA neurons (middle …

https://doi.org/10.7554/eLife.03697.010
Author response image 1

A. Power calculation showing required sample size as a function of effect size to demonstrate significance at p<0.05 level with a power>0.8 assuming log‐normal distributions and a CV of 0.88 (from …

Author response image 2

Strengthening and pruning of HVC‐RA synapses leads to reduced variability, and this trend is robust to changes in either LMAN input strength (A) or intrinsic properties or RA neurons (B). Both …

Author response image 3

Strengthening and pruning of HVC input to RA causes a reduction in variability regardless of whether LMAN firing is Poisson, Poisson with added Poisson bursts, or time varying. A. Average pair‐wise …

Author response image 4

The effect of varying decay kinetics of NMDA currents in our model results in modest changes to RA variability, with longer time constants producing more stereotyped firing.

Tables

Table 1

HVC–RA synaptic properties

https://doi.org/10.7554/eLife.03697.006
BirdsAge (dph)Input resistance (MΩ)Capacitance (pF)Spontaneous firing rate (Hz)SF peak amplitude (pA)SF CVSF latency to peak (ms)MAX peak amplitude (nA)MAX CVMAX latency to peak (ms)
Subsong Juvenile (24)40–45135.84 ± 95.0996.80 ± 22.776.74 ± 4.7929.52 ± 2.580.30 ± 0.219.24 ± 2.940.53 ± 0.260.05 ± 0.048.57 ± 2.13
(55)(55)(55)(72)(72)(72)(29)(29)(29)
Plastic-song Juvenile (23)60–65123.95 ± 93.2695.48 ± 27.477.25 ± 5.1049.60 ± 3.920.19 ± 0.148.39 ± 2.541.31 ± 0.610.05 ± 0.047.47 ± 1.50
(42)(42)(42)(60)(60)(60)(24)(24)(24)
*p < 0.001*p < 0.001**p < 0.001**p < 0.05
Crystalized-song adult (39)90–130105.80 ± 76.0586.71 ± 24.927.95 ± 5.5073.56 ± 5.490.18 ± 0.157.68 ± 2.840.80 ± 0.430.05 ± 0.047.08 ± 1.70
(80)(80)(80)(120)(120)(120)(52)(52)(52)
*p < 0.05**p < 0.001*p < 0.001**p < 0.001**p < 0.001**p < 0.01*
**p < 0.05**p < 0.001*
  1. Values are mean ± SD *Versus Subsong Juvenile; **versus plastic-song juvenile; statistically significant differences in bold.

  2. *

    Two-tail Student's t Test.

  3. Wilcoxon Rank–Sum Test: used when one or more of the distributions under comparison were significantly non-parametric, as determined by the Kolmogorov–Smirnov.

Table 2

LMAN-RA synaptic properties

https://doi.org/10.7554/eLife.03697.008
BirdsAge (dph)SF peak amplitude Vh = +40 mV (pA)SF CV Vh = +40 mVSF latency to peak Vh = +40 mV (ms)Ratio SF peak amplitude at Vh = −70 mV to that at Vh = +40 mVMAX peak amplitude Vh = +40 mV (nA)MAX CV Vh = +40 mVMAX latency to peak Vh = +40 mV (ms)Ratio MAX peak amplitude at Vh = −70 mV to that at Vh = +40 mV
Subsong Juvenile (7)40–45191.30 ± 162.240.10 ± 0.0611.06 ± 3.900.17 ± 0.090.42 ± 0.210.08 ± 0.0412.97 ± 1.710.32 ± 0.28
(40)(40)(40)(40)(18)(18)(18)(18)
Plastic-song Juvenile (13)60–65147.41 ± 132.100.12 ± 0.0510.66 ± 4.290.10 ± 0.090.48 ± 0.280.07 ± 0.0511.56 ± 0.730.28 ± 0.21
(45)(45)(45)(45)(18)(18)(18)(18)
*p < 0.01*
Crystalized-song adult (12)90–130141.24 ± 126.400.10 ± 0.0610.18 ± 4.200.11 ± 0.080.39 ± 0.170.05 ± 0.0311.23 ± 3.380.30 ± 0.17
(38)(38)(38)(38)(15)(15)(15)(15)
*p < 0.001**p < 0.01*
  1. Values are mean ± SD *Versus Subsong Juvenile; **Versus plastic-song juvenile; statistically significant differences in bold.

  2. *

    Two-tail Student's t Test.

  3. Wilcoxon Rank–Sum Test: used when one or more of the distributions under comparison were significantly non-parametric, as determined by the Kolmogorov–Smirnov test.

Download links