Neural activity in cortico-basal ganglia circuits of juvenile songbirds encodes performance during goal-directed learning

  1. Jennifer M Achiro
  2. John Shen
  3. Sarah W Bottjer  Is a corresponding author
  1. University of Southern California, United States
10 figures, 2 tables and 1 additional file

Figures

Figure 1 with 3 supplements
Cortico-basal ganglia circuits for vocal learning in juvenile zebra finches.

Left: core (c, gray) and shell (s, red) subregions of the cortical nucleus LMAN give rise to parallel recurrent loops through the basal ganglia. LMAN-core projects to vocal motor cortex (RA); this …

https://doi.org/10.7554/eLife.26973.002
Figure 1—figure supplement 1
Lesions of AId prevent vocal learning in juvenile birds.

Imitation of tutor (father) songs by adult birds that had received a bilateral lesion of AId as juveniles (at 45 dph, after tutor song memorization) compared to birds that received control lesions. …

https://doi.org/10.7554/eLife.26973.003
Figure 1—figure supplement 2
Distinct populations of neurons in LMAN-shell respond to either tutor song or self-generated song (own song).

LMAN neurons respond differentially to playback of songs in juvenile zebra finches during early stages of sensorimotor integration, following memorization of the tutor song (45 dph). Individual core

https://doi.org/10.7554/eLife.26973.004
Figure 1—figure supplement 3
Individual LMAN-core neurons send axon collaterals into both RA and AId only in juvenile birds.

Left: the borders of AId are difficult to discern in Nissl-stained sections but are clearly demarcated by the axonal projection from LMAN. Labeled axons in this photomicrograph were produced by a …

https://doi.org/10.7554/eLife.26973.005
Figure 2 with 2 supplements
Single neurons in both core and shell subregions of LMAN showed singing-related activity in juvenile songbirds.

Left: Examples of two different core neurons during singing in juvenile birds showing either excitation (top; 54 dph) or suppression (bottom; 43 dph) compared with quiet baseline periods. …

https://doi.org/10.7554/eLife.26973.006
Figure 2—figure supplement 1
Spike bursts increased in excited but not suppressed neurons during singing.

Percent of spikes that occurred in bursts (interspike intervals <10 ms) from core (gray) and shell (red) neurons during singing and local baselines (average of the two baseline periods nearest in …

https://doi.org/10.7554/eLife.26973.007
Figure 2—figure supplement 2
LMAN neurons had low selectivity for different syllable types.

(A) Example spiking responses during production of four different syllable types in a 53 dph bird. Top row shows spectrograms; raster plots and PSTHs for each syllable in single core and shell

https://doi.org/10.7554/eLife.26973.008
Pre-singing activity aligned to syllable onsets showed coordinated premotor activity in core but not shell.

Pre-singing activity in core (gray, top panel) and shell (red, bottom panel) in juvenile birds for all neurons that showed significant excitation prior to syllable onsets. Solid lines show smoothed …

https://doi.org/10.7554/eLife.26973.010
Figure 3—source data 1

Pre-singing spiking activity of individual CORE and SHELL neurons.

https://doi.org/10.7554/eLife.26973.011
Single LMAN neurons encoded similarity to tutor song in singing juvenile birds.

(A) Examples of juvenile syllables with relatively high or low similarity to tutor syllables. Spectrograms (frequency, 0–8 kHz, over time) showing two different tutor syllables (top) and examples of …

https://doi.org/10.7554/eLife.26973.012
Similarity to tutor syllables modulated firing rate in either a positive or negative direction across the population of LMAN neurons.

Standardized response strength for each neuron in core (gray) and shell (red) during production of syllable renditions representing low versus high similarity to corresponding tutor syllables based …

https://doi.org/10.7554/eLife.26973.014
Variability of firing rate was higher during syllable renditions with low tutor similarity in both core and shell neurons.

(A) Coefficient of variation (CV) of firing rate for core (gray) and shell (red) neurons during production of syllables that had high or low similarity to tutor syllables (top versus bottom 50% of …

https://doi.org/10.7554/eLife.26973.015
Figure 7 with 1 supplement
Subsets of syllable utterances during early sensorimotor learning had either high or low similarity to tutor syllables.

(A) Average similarity of all juvenile syllable types to corresponding tutor syllables as a function of the progression of song development from subsong to plastic song (goodness-of-fit coefficients …

https://doi.org/10.7554/eLife.26973.016
Figure 7—figure supplement 1
Measuring degree of song development for each day of singing for each bird.

(A) Left panel: example spectrograms of syllables from the same bird at different ages (47, 49, 50 and 51 dph). Right panel: distributions of syllable durations per day for this bird; red lines …

https://doi.org/10.7554/eLife.26973.017
The association between firing rate and tutor similarity decreased in strength with the progression of song development.

The correlation between baseline-corrected firing rate and tutor similarity (r values, y-axis) is plotted against degree of song development (goodness-of-fit coefficients, x-axis) for neurons with a …

https://doi.org/10.7554/eLife.26973.018
Variability of firing rate decreased in shell neurons during development for syllable renditions with low tutor similarity during subsong.

Top panels: correlation of CV of firing rate during production of syllable renditions in the bottom 50% of tutor similarity with the progression of song development. Bottom panels: correlation of CV …

https://doi.org/10.7554/eLife.26973.019
Figure 10 with 1 supplement
Multiple acoustic features were used to cluster syllable types.

(A) Examples of the features used to calculate acoustic similarity in order to assign juvenile syllable renditions to types (clusters). Top row shows spectrograms for five renditions of a syllable …

https://doi.org/10.7554/eLife.26973.020
Figure 10—figure supplement 1
Examples of syllable renditions with either high or low acoustic similarity to the closest-matching tutor syllable.

The tutor syllable was selected by finding the closest acoustic distance between each juvenile syllable rendition and the tutor syllables for that bird (see Materials and methods).

https://doi.org/10.7554/eLife.26973.021

Tables

Table 1
Response strength during episodes of singing.

Standardized response strength (mean ± s.e.m.) for core and shell neurons in LMAN that showed significant excitation or suppression during song production compared with quiet baseline periods (see …

https://doi.org/10.7554/eLife.26973.009
CORESHELL
FractionResponse strengthFractionResponse strength
Excited0.72 (66/92)7.06 ± 0.710.65 (66/102)7.28 ± 0.44
Suppressed0.28 (26/92)−7.32 ± 1.150.35 (36/102)−5.82 ± 0.45
Table 2
Tutor similarity modulates baseline-corrected firing rates in single LMAN neurons.

Single neurons showed either positive or negative slopes for the regression of firing rate against tutor similarity. The incidence of neurons across the population that had either positive …

https://doi.org/10.7554/eLife.26973.013
Positive slope (r > 0)Negative slope (r < 0)
CORE4850Total cell number (n = 98)
49.051.0Percent
0.082−0.065Mean of r value across all cells
3.32.2Estimated % significant core cells = 5.5
0.22−0.21Approximate mean r value for significant cells (n = 5)
SHELL6260Total cell number (n = 122)
50.849.2Percent
0.062−0.081Mean of r value across all cells
4.56.3Estimated % significant shell cells = 10.8
0.21−0.31Approximate mean r value for significant cells (n = 13)

Additional files

Download links