Zebra Finch and Mental Illness in Humans

This article is based on the paper entitled “Incomplete and Inaccurate Vocal Imitation after Knockdown of FoxP2 in Songbird Basal Ganglia Nucleus Area X” written by Sebastian Haesler, Christelle Rochefort, Benjamin Georgi, Pawel Licznerski, Pavel Osten, and Constance Scharffand.  This article was published in PLOS Biology, Volume 5 Issue 12 in December 2007.


Imagine not being able to communicate your thoughts or emotions.  Human language is a key aspect of the way we communicate throughout our lives. Developmental verbal dyspraxia (DVD) is a disorder that limits this ability to communicate by causing various language deficits.  Forkhead box transcription factor (FoxP2) is an essential protein for complete language development in humans and, when mutated, causes DVD in humans.  The key affected region has been revealed to be the portion of the brain known as the basal ganglia.  Area X is the basal ganglia homologue in the brains of songbirds and is a key aspect to their song development (song development to birds is like speech development in humans).  Studies have shown that human speech development and learned song vocalization in birds have many commonalities (Doupe and Kuhl, 1999). This makes songbirds a good fit as a model organism for studying the neural basis of learning and memory. One songbird in particular, the zebra finch (Taeniopygia guttata), has been a key model organism for this kind of research because of its inclination to sing and breed in captivity and its rapid maturation (Fee and Scharff, 2010).

Key Point: Humans and songbirds have parallels in how their communication systems develop on a neural level.  One of these links is the FoxP2 protein which can be studied to further explore these neural mechanisms and their effects.

Part 1: Song Imitation of FoxP2 Knockdown Zebra Finches

  • Goal: To investigate the requirement of FoxP2 for normal song development in the zebra finch, a model for studying the basic principles of vocal learning.
  • Preparation: Lentivirus-mediated RNA interference (RNAi) was used to reduce FoxP2 mRNA and protein levels in Area X. This was done by injecting one of 2 different short interfering hairpin RNA (shRNAs)— shFoxP2-f and shFoxP2-h. Both of these shRNAs are designed to target different sequences in the FoxP2 gene. Two shRNAs were also designed as controls and did not target any zebra finch gene (shControl and shGFP).  Both the control and the experimental shRNAs were tested extensively to confirm that the controls had no effect and that the experimental shRNAs induced specific, long-lasting knockdown of gene expression in zebra finch Area X without causing cell death.
  • Details: After establishing their experimental and control injections, Haesler et al. (2007) compared knockdown and control pupil’s songs to their tutor’s when they reached 90 days after hatching (post-hatch day [PHD]) to see how much of a resemblance they share and, in turn, how much success the pupil has. If reduced FoxP2 levels were to have an effect on learning, one would expect to see a poor song imitation by the knockdown pupils.
  • Results:

Figure 1 (left) Incomplete Tutor Song Imitation By FoxP2 Knockout Pupils.

(a) Tutor and control songs are the same— compare the patterns. (b & c) Each of the two FoxP2 knockouts renditions are not the same as their tutors.  (d) The mean similarity scores between tutor and pupil motifs were significantly lower in knockdown pupils than in either of the control-injected pupils. There was no significant difference between the shControl- and shGFP-injected birds.

Figure 2 (Right) Inaccurate Tutor Song Imitation By FoxP2 Knockout Pupils.

(a) Tutor and control bird songs are the same, while the knockout is different. Red italic letters denote imprecisely copied syllables. (b) The average accuracy was significantly lower in knockout pupils when compared to controls. (c) Shows the identity scores of FoxP2 knockouts in dark grey having a lower average score when compared to the control in light grey. This suggests that all syllables are affected. (d) Compares precision of syllable duration and acoustic value. There was a larger divergence of syllable imitation for all measures in the knockouts shown in dark grey than the controls shown in light grey.

Part 1 Take Home Message: These data strongly suggest that insufficient FoxP2 levels result in both incomplete AND inaccurate imitation of the song, implicating FoxP2 in postnatal brain function.

Part 2: Song Performance and Development in FoxP2 Knockdown Zebra Finches

  • Goal: The incomplete and inaccurate vocal imitation of tutor song in FoxP2 knockdown pupils raises the question whether knockdown pupils were unable to generate particular sounds. To investigate the variability of song syllables in FoxP2 knockdown and control zebra finches.
  • Details: Using the same experimental and control injections as before, Haesler et al. (2007) first, compared sonograms from different renditions of the same syllable in each of the birds at PHD90. Next, they quantified the variability of syllable duration between different renditions of the same syllable. Finally, Haesler et al. (2007) analyzed the sequential order of syllables over the course of many motifs (a motif is just syllables that are rendered in a stereotyped sequential order).
  • Results:

Figure 3 Variability of Syllable Production in FoxP2 Knockdown Pupils.

(a & b) The comparison the same syllable between different birds revealed that knockdown pupils sang their syllables more variably than control pupils. Each column is a different syllable. Compare the line patterns of each rendition. (c) Acoustic variability of syllables from rendition to rendition was higher in knockdown pupils than in control pupils (shGFP and shControl injections). The controls and tutors performed their syllables with equal stability. (d) Syllable duration varied more from rendition to rendition in knockdown pupils (shFoxP2) than in controls (shControl and shGFP) and tutors. There was no difference between the tutors or either of the control birds.

Asking the Right Questions: Haesler et al. (2007) expanded this question a bit by analyzing the songs of knockdown and control pupils over various stages of development—PHD60, 80, and 90. The PHD90 data is from the earlier trials.


Figure 4 Differences in Song Development of FoxP2 Knockdown and Control Pupils

(a) This graph shows accuracy score. Knockdown pupils were already imitating their tutors less accurately than control pupils at PHD65. Additionally, knockdown pupils did not improve in accuracy after PHD80, unlike the control pupils.  This suggests that the knockdown pupils had reached the end of their learning phase.

(b) This shows accuracy variance values between knockdown and control pupils to look at variability of syllable production during song development. The variance was similar between the experimental and control groups at PHD65 and at PHD80, but was significantly higher in knockdown pupils compared to controls at PHD90.

Part 2 Take Home Message: Knockdown pupils sing as variable as control pupils during early song development and even more variable as adults; this made Haesler et al. (2007) favor the opinion that reduced FoxP2 levels compromised the knockdown pupil’s ability to match their motor output to the memorized model of the song the pupil has from the tutor from learning.

Overall Take Home Message

1) These results bridge more similarities between human speech and song development in songbirds at a molecular level, further validating songbirds as a strong model organism for neural studies of speech.

2) We saw that a reduction of FoxP2 has an impact on both song learning and speech development. This supports the hypothesis that during evolution, the ancestral genes and neural systems in humans adapted to provide a distinct language ability.

3) These results detail functional gene analysis in songbirds which suggest that FoxP2 is mandatory for normal auditory-guided vocal motor learning.

Strengths and Weaknesses in This Paper

(1) Weakness: Might have been better to use the same n for all of the samples in Figure 1 (GFP control has 3, others have 7).

(2) Strength: Haesler et al. (2007) used a Mann-Whittney U test for pretty much all of their analyses. This was the proper choice of statistical method because there was always a “k” of 2 with non-parametric, independent data. Additionally, they were sure to correct for alpha in their calculations when appropriate (the probability of committing type I error, aka a rejecting a true null hypothesis).

(3) Strength: Haesler et al. (2007) did a good job of using controls and doing pilot testing to confirm that they were appropriate. They also took into account aspects that might have been overlooked by other authors. Some examples are,

  1. They verified in vitro that knockdown of FoxP2 did not affect protein levels of FoxP1, the closest homolog of FoxP2.
  2. Haesler et al. (2007) confirmed that FoxP2 knockdown did not cause apoptotic cell death in Area X and that it did not alter the density of neurons in this nucleus.
  3. Additionally they ran tests to eliminate the possibility that unspecific effects of RNAi induction, viral infection, or damage to Area X influenced their results. Additionally, Haesler et al. (2007) made sure that specific song features of the pupil’s tutor did not contribute to behavioral differences.

(4) Weakness: Needs a future study to look isolate impairments in motor production and motor learning, because song analysis can’t be the sole factor in this determination.

Extra Media and Info

If you are curious, here is a video of an attempted zebra finch courtship

If you would like to read more details on neural mechanisms of birdsong memory,
Bolhuis, J and Gahr, M (2006) have a good review here titled “neural mechanisms of birdsong memory”

Here is good overall review of zebra finch written by Fee, M and Scharff, C (2010) titled “The Songbird as a Model for the Generation and Learning of Complex Sequential Behaviors”


Doupe, AJ & Kuhl, PK. 1999. Birdsong and human speech: common themes and mechanisms. Annual Review of Neuroscience 22:567–631.

Fee, M & Scharff, C. 2010. The Songbird as a Model for the Generation and Learning of Complex Sequential Behaviors. ILAR Journal 51:362-377

Haesler S, Rochefort C, Georgi B, Licznerski P, Osten P, Scharff C. 2007. Incomplete and inaccurate vocal imitation after knockdown of FoxP2 in songbird basal ganglia nucleus Area X. PLoS Biol 5:e321.

Leave a Reply

Your email address will not be published. Required fields are marked *