More on this book
Kindle Notes & Highlights
kinematic variables indicate the desired outcome for the arm,
deal with hyper-planes in a fourteen-dimensional space, visualise a 3-D space and say “fourteen” to yourself very loudly.’
brain needs to balance efficient information transmission with the need to be robust and so should have some redundancy.
While most physical networks like roads or telephone lines will never be built based on pruning, digital networks that don’t have costs associated with constructing edges
Whereas regular neurotransmission is like a letter sent between two neurons, neuromodulation is a leaflet sent out to the whole community.
diversity doesn’t always mean
difference. What may look like a deviation from the structural norm could in fact be a perfectly valid way to achieve the same outcomes.
Kant’s philosophy, the ‘Ding an sich’, or ‘thing-in-itself’, refers to the real objects out in the world – objects that can’t be experienced directly, but only through the impressions they make on our sense organs.
Taking further inspiration from Kant, Helmholtz proposed that this inference proceeded by interpreting the current sensory input in light of pre-existing knowledge about the world.
According to Cardano: ‘Gambling is nothing but fraud and number and luck.’
‘When you hear hoof beats, think of horses, not zebras’ is a bit of advice frequently given
excellent example of the rule of inverse probability in action.
If, however, we find ourselves in a world that deviates from the one we’ve evolved and developed in, our priors can be misleading. ‘Think of horses’ is only good advice in a place with more horses than zebras.
drivers have a tendency to speed up in the fog – with weak information about their own movement, they assume their speed is too slow.
‘This ability to describe the data accurately comes at the cost of falsifiability.’
humans actually assume that light comes from above and slightly to the left
The statistician and historian responsible for this assessment, Stephen Stigler, is also known for ‘Stigler’s law of eponymy’, which claims that a scientific law is never named after its true originator. The sociologist Robert Merton is believed to be the originator of this law.
Sutton’s algorithm – known as ‘temporal difference learning’ – beliefs are updated in response to any violation of expectations.
‘learning a guess from a guess’ as Sutton describes it.
connection if it happens in the presence of dopamine. Dopamine – which encodes the error signal needed for updating values – is thus
also required for the physical changes needed for updating that occur at the synapse. In this way, dopamine truly does act as a lubricant for learning.
understand any bit of the brain we should be able to explain it on each of three levels: computational, algorithmic and implementational.
The computational level asks what is the overall purpose of this system, that is, what is it trying to do? The algorithmic level asks how, i.e., through what steps, does it achieve this goal. And finally, the implementational level asks specifically what bits of the system – what neurons, neurotransmitters, etc. – carry out these steps.
‘That’s interesting!’ In it, he said: ‘It has long been thought that a theorist is considered great because his theories are true, but this is false. A theorist is considered great, not because his theories are true, but because they are interesting … In fact, the truth of a theory has very little
to do with its impact, for a theory can continue to be found interesting even though its truth is disputed – even refuted!’
Thousand Brains Theory is a piece of neuro-architecture known as the cortical column.
They’re so-named because they form cylinders running from the top of the neocortex through to the bottom,
the neurons are segregated into six visibly identifiable layers.
Typically, all the neurons in a column perform a similar function; they may all, for example, respond in a similar way to
Yet the different layers do seem to serve some different purposes: some layers, for example, get input from other brain area...
This highlight has been truncated due to consecutive passage length restrictions.
fundamental anatomical unit of the brain.
Grid cells are neurons that are active when an animal is in certain special locations. Imagine a mouse running around an open field. One of its grid cells will be active when the mouse is right in the middle of the field. That same cell will also be active when the mouse has moved a few body lengths north of the centre and then again a few lengths north of that.
There is a problem, however: grid cells aren’t found in the neocortex.
At a glance the neocortex may appear like a uniform tessellation,
Almost all competing theories of consciousness, it seems to me, have been so vague, fluffy and malleable that they can only aspire to wrongness.’
