Based on information-transfer statistics for humans, once one becomes adept at a language, irrespective of the language, the information transfer rate is about 40 bits per second (which translates into a trillion possibilities per second, Coupé et al. 2019). But to learn a language, the transfer rate is many orders of magnitude lower. A group of Japanese university students, who were moderately bilingual, were enrolled in a 4-month intensive language course to improve their English (Hosoda et al. 2013). During this period, they learned ~ 1000 new English words which they used in various spoken and written contexts. The learning was followed by a weekly test. To learn 1000 words, it is estimated that 0.0006 bits per second of information were transmitted for storage over the 4-month period [1.5 bits per letter x 4 letters/word x 1000 words/16 weeks, using the method of Reed and Durlach 1998]. Thus, the rate of transfer for learning a language is over 10,000 times lower than the rate of transfer for executing a language once mastered [the value for children learning a first language is comparable at 0.0008 bits per second from birth to the age of 18 to achieve a vocabulary of 60,000 words; Bloom and Markson 1998; Miller 1996].
This difference in the rate to store learned information and the rate to execute a response explains why a large number of neurons have been dedicated to the acquisition of language, which requires a functional hippocampus, neocortex, and cerebellum (Corkin 2002; Kimura 1993; Ojemann 1991; Penfield and Roberts 1966; Schmahmann 1997), and which together represent some 85 billion neurons (Herculano-Houzel 2009). Of course, not all neurons are dedicated to language, but it has been estimated that just under half of the neurons in the brain subserve this function (Sarubbo, Duffau 2020). To automate a behavior including language processing, it cannot be done without the cerebellum (Hasanbegović 2024), which converts the declarative conscious code of neocortex into executable code by way of the muscles for speaking, writing, and reading (Tehovnik, Hasanbegović, Chen 2024).
Even though the cerebellum, which contains the most neurons in the brain and which possesses a plethora of Purkinje neurons for finalizing a learned response (Huang 2008), once an automated state has been achieved through learning it is only the cerebellar nuclei (of the cerebellar structure) that seem necessary for task execution once all the synaptic weights have been optimized (see Fig. 1; Ito et al. 1974a; Kassardjian et al. 2005; Miles and Lisberger 1981; Sendhilnathan, Goldberg 2020b; Takahara et al. 2003; Takemori and Cohen 1974). Patient HM, who had bilateral destruction of the hippocampus, is a case in point (Corkin 2002). Even though he was able to communicate using the English language, he would not be able to acquire a second language, which necessitates the long-term storage of linguistic information in the neocortex (Ojemann 1991; Penfield and Roberts 1966) and it requires the simultaneous conversion of this information into executable code by the cerebellum for speaking, writing, and reading in the second language.
Figure 1: This figure illustrates a minimal circuit for the creation and execution of a learned response, the vestibulo-ocular reflex. If the gain of the reflex needs to be changed because magnifying or minimizing prisms have distorted the visual world of a subject, then Purkinje neurons are summoned to modify the gain. Once modified, this circuit is no longer necessary to maintain the automated response and a circuit bypassing the Purkinje neurons via the cerebellar nuclei (i.e., the vestibular nuclei) is sufficient to maintain the response (Miles and Lisberger 1981). This principle applies to all learned behaviors (Tehovnik, Hasanbegović, Chen 2024). Illustration from Lisberger and Fuchs (1978).
- Badges
- Science topic
- Similar topics
- Political Science
- International History and Politics
- Globalization