From bird songs to frog ribbits, animals engage in countless forms of vocalization. However, no other species in the animal kingdom matches humans in complexity of language. The versatility of human speech allows us to discuss anything from what we ate for breakfast to the nature of the universe, and our ability to communicate is essential in all aspects of our lives. Because of this, it is natural for neuroscientists to search for an evolutionary explanation showing us how our unique language capabilities came about. One potential answer to this complicated question lies in the gene FOXP2.
The importance of FOXP2 in speech is perhaps best underscored by the fact that individuals with FOXP2 mutations have severe deficits in speech production. The fact that such severe speech abnormalities result from a single gene has led neuroscientists and evolutionary biologists alike to extensively study FOXP2 in hopes of understanding the origin of human speech. FOXP2 is a transcription factor, meaning that it regulates the expression levels of other genes. This allows FOXP2 to regulate the synthesis of a wide range of other proteins, and differences between the human FOXP2 gene and the FOXP2 gene of other animals may contribute to the complexity of human speech. Curiously, though, researchers have noted that “FOXP2 is among the 5% most-conserved proteins” between humans and rodents. How could such small differences between the FOXP2 of humans and rodents contribute to such a large difference in vocalization?
“Amino acids are like the lego blocks used to build proteins.”
After a closer analysis of the specific differences between the sequence of amino acids in the protein FOXP2 in humans, rodents, and primates, the researchers found an answer to this question. Amino acids are like the lego blocks used to build proteins. While there are only twenty amino acids commonly used in proteins, there are countless ways to order them and each of these possible orderings makes a protein with a unique structure and function. One of the changes in the FOXP2 amino acid sequence that is unique to humans creates a special site called a phosphorylation site for the signaling molecule Protein Kinase C (PKC). This means that when PKC is activated, it is able to add a chemical called a phosphate group to one of the amino acids in FOXP2. Adding phosphate groups to specific amino acids of proteins often alters their functions — in fact, it is one of the most common ways a cell regulates the activity of proteins.
Thus, activation of the PKC signaling pathway may lead to the phosphorylation of FOXP2 and downstream changes in how FOXP2 regulates the production of other proteins. Although this change seems small (after all, it is only one amino acid in a much larger protein) it dramatically alters the ability of cells to regulate all of the proteins under the control of FOXP2. While this study showed how human FOXP2 is different from that of animals, it does not reveal how these changes affect the development of speech.
More recently, researchers have began to wonder how exactly the human-specific FOXP2 mutations affect the brain. To explore this, Christiane Schreiweis, Urlich Bornschein, and their colleagues investigated how FOXP2 affects mouse learning. It’s natural to wonder what mouse learning has to do with human speech, but interestingly enough, it offers quite a bit of insight into how our unique vocal capabilities came about. Studying mice is far simpler than studying humans or primates, both because it is easier to make modifications to their genetic code and because many well-established behavioral tests for mice exist. Furthermore, as there are significant ethical issues related to experimentation—especially gene editing—in humans and nonhuman primates, using a mouse model represented a less controversial alternative. In order to determine what makes human FOXP2 so unique, the researchers prepared a battery of tests to compare wild type mice with mice genetically modified to express the human FOXP2 gene, referred to as “humanized FOXP2” mice. Wild type mice, of course, still possess the mouse FOXP2 gene.
A first step in understanding how FOXP2 influences language is testing how FOXP2 affects learning. The complexity of human language — both in terms of its syntax and the motor skills necessary to produce coherent speech — makes studying the mechanisms underlying language learning especially relevant. However, as mice are unable to learn to speak like us, even with humanized FOXP2, the researchers used alternative experimental approaches to study the effect of FOXP2 on mouse learning.
One important finding of this study was that while mouse motor learning is equally efficient in both groups, humanized mice performed far better than their wild type counterparts in a more complex task known as the conditional T-maze. In this task, the mice are placed at the end of the long arm of the T and forced to decide between which of the two arms to choose in order to receive a food reward. The mice could either remember it through procedural cues (for example, turning towards the side which has a rougher floor) or through declarative cues (for example, turning towards a spatial cue like a star-shaped sticker outside the maze). By manipulating which cues were present, the researchers were able to examine the relationship between procedural and declarative learning. The fact that humanized mice were more successful than wild type mice in this task suggests that FOXP2 reduces competition between these distinct forms of learning. These results, along with those from related experimental tasks, led the researchers to conclude that mice with humanized FOXP2 “exhibit enhanced ability to make transitions from a declarative to a procedural mode of learning.” Instead of having the two forms of learning working at the same time (and to some extent interfering with each other), the mice with humanized FOXP2 quickly began to preferentially activate the procedural learning system; this allowed them to learn complex tasks more quickly than mice with wild type FOXP2. These differences in the learning strategies of mice with humanized and wild type FOXP2 suggest that a more efficient transition to a procedural mode of learning could explain why humans are able to learn to speak far more effectively than other animals.
In addition to studying mouse behavior, the researchers wanted to gain a better understanding of the effects of FOXP2 at the molecular level. Specifically, they looked at the dorsomedial and dorsolateral parts of the striatum, a part of the brain linked to learning. Previous work by other researchers had already established that the dorsomedial striatum is closely linked to declarative learning, while the dorsolateral striatum is related to procedural learning. Based on the behavioral data the researchers had gathered, the researchers hypothesized that there would be altered signaling activity within these regions. The researchers found a statistically significant difference in levels of the neurotransmitter dopamine between mice with humanized and wild-type FOXP2 in the dorsomedial striatum, with the mice expressing humanized FOXP2 showing lower dopamine levels. In the dorsolateral striatum, a nearby brain region, researchers found that long-term depression (LTD), a process through which neural connections are weakened, was stronger in mice with humanized FOXP2 than in wild-type controls.
“…how do these findings on mouse learning inform our understanding of human speech?”
These biochemical findings may explain why the two groups of mice learn differently. Lower levels of dopamine signaling in the dorsomedial striatum in mice with humanized FOXP2 may lower the overall activity in the region, making declarative learning less efficient in mice. With this change, the dorsolateral striatum and procedural learning would be expected to play a larger role in learning; indeed, the researchers observed the dominance of procedural learning in tasks where the two forms of learning competed. Furthermore, LTD can be thought of as the process by which synapses “learn,” and an increased efficacy of this process in the dorsolateral striatum provides additional support to the hypothesis that dorsolateral striatum-mediated procedural learning becomes more effective in mice with humanized FOXP2. Overall, these findings provide an anatomical explanation for how humanized FOXP2 leads mice to favor procedural learning more than mouse FOXP2 does.
Now, how do these findings on mouse learning inform our understanding of human speech? One proposal is that human FOXP2 allows the linking of multiple speech-related actions into a coherent process by favoring procedural learning, and the striatum has been found to be involved with this type of behavior in previous research. In particular, this research is in line with the Declarative/Procedural (DP) Model of language. This model proposes that while our mental vocabulary is stored in declarative memory, verbal articulation and the combination of words into coherent sentences is more closely linked to procedural memory. When we learn to tie our shoes, we are able to use procedural memory to turn a complex series of steps into a single fluid action. FOXP2 may allow us to learn speech in a similar manner. However, it is difficult to attribute a phenomenon as complex as language to a mutation in just one gene. There are certainly many other factors contributing to our unique language capabilities, and in the coming years neuroscientists will undoubtedly be searching for further insights into the evolution of speech.
Feature image by Kayleen Schreiber, diagram by Huixuan Liang.
Enard, W., Przeworski, M., Fisher, S. E., Lai, C. S. L., Wiebe, V., Takashi, K., … Paabo, S.
(2002). Molecular evolution of FOXP2 , a gene involved in speech and language. Nature, 418(August), 869–872.
Schreiweis, C., Bornschein, U., Burguière, E., Kerimoglu, C., Schreiter, S., Dannemann, M., … Graybiel, A. M. (2014). Humanized Foxp2 accelerates learning by enhancing transitions from declarative to procedural performance. Proceedings of the National Academy of Sciences , 111(39), 14253–14258. http://doi.org/10.1073/pnas.1414542111
Ullman, M. T. (2004). Contributions of memory circuits to language: the declarative/procedural model. Cognition, 92(1), 231–270. http://doi.org/http://dx.doi.org/10.1016/j.cognition.2003.10.008