Studying singing birds is an interesting model to understand how language arises and is modulated by the brain. Singing birds modulate their song depending on the social context (undirected when single, directed when in presence of a female). What are the neural pathway modulating this difference of song, and how are the mechanics? This presentation reviews a recent work finding that LMAN may be modulating song variability, purposely to allow to explore new optimal song syllables. This may have an impact in further researches on human children language learning, as variability may be an important mechanism to help learn new letters and syllables.

1. Birds’ song variability is
modulated by LMAN
in anterior forebrain pathway
Hugo Trad & Stephen Larroque
19 Nov 2015

2. Learning to sing
• Birds’ singing is a complex, learnt task
• Uses cortical-basal ganglia circuit
• Different song characteristics between
directed (couple) and undirected (single)
• Song variability : how and why ?
(how to demonstrate direct contribution of AFP to adult
birdsong?)
2
(Kao et al, 2005)

3. AFP Pathway
3
(Kao et al, 2005)
Motor
Pathway
Basal ganglia-forebrain circuit
(anterior forebrain pathway)

4. What we know
• AFP activity correlated with singing
• LMAN lesion → no change in ongoing song
• HVC+RA (motor pathway) stimulation → vocalizations
• LMAN stimulation → no vocalization elicited and no gross
change on song
=> LMAN not required structure for song production
=> What is its purpose then ? Hypothesis :
AFP activity modulates vocal control via LMAN activity
patterns ?
4
(Kao et al, 2005)

5. Methodology: manipulating LMAN activity
5
(Kao et al, 2005)
Microstimulation of LMAN during precisely targeted parts
of a song using real-time template-matching algorithm

6. Results: pitch+loudness modulated by LMAN
6
(Kao et al, 2005)
(sd ; P < 0.0001) (sem ; P < 0.0001)
Frequency shift Amplitude shift
Using fixed current stimulation

7. Results: LMAN is functionally compartmentalized
7
(Kao et al, 2005)
And high current produce song suspension, like HVC stimulation !

8. Methodology: social behavioral analysis
8
(Kao et al, 2005)
Spectrogram + neural activity recordings when presented
with a female
Undirected singing Directed singing

9. Methodology: social behavioral analysis
9
(Kao et al, 2005)
Spectrogram + neural activity recordings when presented
with a female
Undirected singing Directed singing

10. Methodology: social behavioral analysis
10
(Kao et al, 2005)
Spectrogram + neural activity recordings when presented
with a female
Undirected singing Directed singing
After <= 2 min, hide female + play undirected songs

11. Results: Context modulates LMAN activity
11
(Kao et al, 2005)
Pitch mean is not changed (needs confirmation...), only variability !
(mean
P = 0.54)
(variance
P < 0.0001)

12. Correlation of LMAN activity and song variability
12
(Kao et al, 2005)
=> LMAN necessary for context-dependent variability
(confirmed by stereotaxic lesion)
(R² = 0.83 ; P < 0.05) (s.e.m. ; pre-P < 0.0001 ; post-P = 0.167)

13. Discussion
• What pathway(s) for context modulation ? (what modulates
LMAN difference in activity between directed and undirected
singing? Dopamine ?)
• Exploration-exploitation dilemma : what is the exploration
probability ? And how is this value chosen ? Is it dynamic ?
• How are those new explored syllabes evaluated ?
• Cerebellar implication ? (to bias variability towards the
correct pitch in juvenile learning ?)
• Different sample sizes for the two conditions (class skewing)
• Need extended theory about amplitude modulation (here the
article is mainly studying frequency!) 13

14. Take home message
• LMAN induces acute changes in individual song
elements (syllabes) frequency (pitch) and amplitude
(loudness), without altering order nor structure of
subsequent syllabes
• Noise and variability can be used advantageously to
explore new unknown solutions, but can become
pathological (Parkison?)
• Exploration behavior promoted in undirected singing
(« training mode »), while exploitation in directed
• Natural organisms exhibit a solution to the exploration-
exploitation dilemma close to the epsilon (noise) trick
→ possible predictive model ? 14

15. – Contributions of an avian basal ganglia–forebrain circuit to real-time
modulation of song, Kao, Mimi H., Allison J. Doupe, and Michael S.
Brainard, 2005 Nature 433.7026 (2005): 638-643.
– Behavioral and Neural Signatures of Readiness to Initiate a Learned
Motor Sequence, Rajan R. & Doupe A., 2013, Current Biology, 23 (1) 87-93.
– Interruption of a basal ganglia–forebrain circuit prevents plasticity of
learned vocalizations, Brainard, M. S., & Doupe, A. J., 2000, Nature,
404(6779), 762-766.
– Social context modulates singing-related neural activity in the songbird
forebrain, Hessler, N. A., & Doupe, A. J.,1999, Nature neuroscience, 2(3),
209-211.
References
slideshare.net/LRQ3000

16. Thank you!
slideshare.net/LRQ3000

17. Bonus Slides

18. THE END