In the dystopian world of George Orwell’s Nineteen Eighty-Four, the government of Oceania aims to achieve thought control through the restriction of language. As explained by the character Syme, a lexicologist who is working to replace the English language with the greatly-simplified “Newspeak”: “Don’t you see that the whole aim of Newspeak is to narrow the range of thought?” While Syme’s own reflections were short-lived, the merits of his argument were not: the words and structure of a language can influence the thoughts and decisions of its speakers. This holds for English and Greek, Inuktitut and Newspeak. It also may hold for the neural code, the basic electrical vocabulary of the neurons in the brain.
Neural codes, like spoken languages, are tasked with conveying all manner of information. Some of this information is immediately required for survival; other information has a less acute use. To accommodate these different needs, a balance is struck between the richness of information being transferred and the speed or reliability with which it is transferred. Where the balance is set depends on context. In the example of language, the mention of the movie Jaws at a dinner party might result in a ranging and patient—if disconcerting—discussion around the emotional impact of the film. In contrast, the observation of a dorsal fin breaking through the surf at the beach would probably elicit a single word, screamed by many beachgoers at once: “shark!” In one context, the language used has been optimized for richness; in the other, for speed and reliability.
In the brain, this same type of balance is usually thought to be an effect of the division of labor. Certain regions—for example the cingulate cortex—are involved in processing higher-level emotional and motivational information. Other regions, like the amygdala, work to keep you safe from more immediate dangers. In other words, one helps you at the dinner party, and the other at the sea. These specialized functions have often been attributed to anatomy: one region might have greater or fewer neurons than the other, and those neurons might wire into different circuits. In either case, it has been assumed, the neurons present are using neural codes with the same basic design.
Neurons can either fire or remain silent, and the combination of the two over time gives rise to a neural code, like dots and dashes in Morse code. As with Morse code, there are theoretical limits on the richness and speed of information transfer. A new Morse code with a thousand characters could exchange richer information, but the speed and reliability of its SOS signal would suffer. Neural codes accommodate this trade-off in their design, and it has been presumed that from neuron to neuron and region to region the balance between richness and speed is the same.
But a closer look at what exactly the neurons in the human cingulate cortex and amygdala are saying has revealed that they employ strikingly different neural codes. One is optimized for richness, and one for speed—just such a trade-off as might be expected given the function of these brain regions. Moreover, a comparison of human and monkey brains has revealed that in both of the studied brain regions, the code used by human neurons is more rich. In effect, different regions—and the brains of different animals—use different neural codes.
These discoveries, published earlier this year in Cell, have wide-ranging and potentially stunning implications. The function of a neural circuit—whether it underlies echolocation, feeding, or any other behavior—is often understood by its wiring diagrams. As with an electrical diagram, many pieces are considered to be interchangeable—a resistor is a resistor and a switch is a switch. Thus, a circuit diagram made up of mouse, monkey or human neurons might be expected to perform the same computation. These new findings challenge that idea, showing that even the basic building blocks in two regions of the same brain can behave very differently.
It is as though some regions of the brain employ an English vocabulary; and others employ that of Newspeak.
Given the interregion and interspecies nature of this study, its results will require plenty of additional corroboration and support before they can be fully adopted and generalized. But at a minimum, the authors have shown the following: in both humans and rhesus monkeys, neurons in the cingulate cortex employ a richer neural code than neurons in the amygdala. Not only is the code simpler in the amygdala, but multiple amygdala neurons were often observed to use the same code in concert. As in the example of the shark at the beach, this is thought to aid in the robust and rapid transmission of information about immediate threats. Finally, neural codes are more rich in human brains than in macaque brains, regardless of the region. In total, these findings suggest that there is a tradeoff between regions and species in the neural code. This tradeoff likely helps to shape the cognitive or computational capacities of different brains and different brain regions.
More broadly, this study highlights a number of routes for additional work. For starters, while many diseases and disorders of the brain have obvious physical manifestations that can be detected with an x-ray or an MRI, some do not. Changes to the neural code, even in the absence of another injury or insult, could therefore be an important but undiscovered driver in psychiatric disease. Perhaps even more important are the implications for the pursuit of research at large. Biology is a field that relies heavily on the transferability of its results: one sentence in a biology textbook may have been derived from experiments in different animals and different contexts. Understanding which pieces of the biological puzzle are interchangeable and which are not is indispensable for the construction of knowledge paradigms.
On the horizon is another interesting future direction. The use of neural code-based comparative approaches like those laid out by the authors might allow for a less biased measurement of the cognitive capacities of other animals—work that could ultimately aid in conservation efforts. Why do we need such an approach? Because we cannot understand intelligences that are unlike our own. But that doesn’t mean they don’t exist. To borrow a word from Orwell, some of them might even be doubleplusgood.