Single-cell organisms can be conditioned (Saigusa et al. 2008); therefore, it should be expected that single cells in the neocortex can also be conditioned (Prsa et al. 2017). In the study of Prsa et al. (2017), a cell was conditioned in the motor cortex of a mouse (as evidenced by two-photon imaging) and feedback of successful conditioning was achieved by optogenetic activation of cells in the somatosensory cortex. A mouse (with head-fixed) was rewarded with a drop of water following volitional discharge of a motor cell using the method of Fetz (1969). The conditioning was achieved after 5 minutes of practice. Furthermore, a group of three cells was conditioned such that two cells were made to fire at a high rate and one cell was made to fire at a low rate, which indicates the inherent plasticity of the nervous system.
This is the first example of the brain having single-cell resolution for transferring information. It is thus not surprising that the brain of a human being (which contains ~ 100 billion neurons) can transfer 40 bits per second (over a trillion possibilities per second, i.e. 2^40 per second) when engaged in language execution (Reed and Durlach 1998), but only after many years of training. If we assume that each cell in the human brain has (on average) at least 10 levels of firing-frequency, then 100 billion neurons should be able to transfer 1 trillion output possibilities (i.e. 10 x 100 billion) or about 40 bits of information, and all done in a second.
And to free up memory space for running the heart and lungs we have information chunking (Miller 1956), so that a concept like ‘E = mc^2’ (as developed by Einstein) can be memorized and used to extract pertinent information as stored in any physics library (Clark 1998; Varela et al. 1991). The availability of books following the development of the printing press in 1436 (by Johannes Gutenberg) has contributed to world literacy by amplifying the information available to the human brain. Artificial intelligence will, no doubt, further enhance the amplification. In fact, much of what I have written over the years has been supported by Google/ResearchGate/AI—and this is without ever using chat-GPT to compose a text.
Dear Moonlit Melody,
I would suggest that using single-celled organisms to study Parkinsonism will not work since Parkinson’s is an attribute of multicellular animals that use dopamine to initiate ‘drive’. However, single-called organisms could be used to study the genetic basis of learning and memory, as was done by Eric Kandel (2006) who developed methods to investigate such things in the mammalian nervous system. I suspect that a comparable genetic mechanism for learning would be found in the single-cell case, and if not this would be an interesting topic for evolutionists.
As for using biology as a metaphor to understand how AI works, Elon Musk has made predictions of how his brain implants will accelerated the amount of information transmitted from the current average level of 10 bits per second (1,024 possibilities, 2^10) to over a million bits per second. Plunging a bunch of wires into the brain willy-nilly is about as useful as expecting someone who does not understand a foreign language to extract meaning from that language. Even the cochlear implant, which changes the way one’s language sounds, requires over one year of training—and music appreciation is never fully recovered. In short, the brain has to be treated like a system composed of functional modules before meaning can be extracted from its neurons and synapses.
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity