Fig. 3. Distributions of response latencies relative to syllable onset at different stages of vocal development.

a An example putative NIf_HVC neuron recorded during subsong (49 dph), the earliest stage of vocal development. b Another example of a NIf neuron recorded during subsong (47 dph). Note the premotor latencies relative to syllable onset. c Histogram of latency relative to syllable onset for song-locked putative NIf_HVC neurons at subsong (50 dph or younger), protosyllable (51–59 dph), and multi-syllable (60 dph or older) stages. Latency is defined as time of the peak in the syllable-onset-aligned PSTH. Triangles indicate median. d For comparison, histogram of latency relative to syllable onset for HVC projection neurons across development, reproduced from ref. ⁴⁰. Here, song stages are defined based on syllable duration distributions, with median and interquartile range 48 ± 4 dph for subsong; 58 ± 10 dph for protosyllable song and 62 ± 12 dph for multi-syllable stage (n = 19, 104, and 814 neurons, respectively). e Schematic of possible HVC network connectivity at each stage of development⁴⁰, consistent with latencies observed in HVC and NIf. Neurons receiving external input are marked by filled circles, with different colors representing syllable-specific inputs. Connections between neurons are marked by arrows, where darker arrows represent stronger connections.