I thought that always as a strange argument. Why would relative size matter for intelligence? At least if one thinks of compute capacity this shouldn't matter.
Both absolute size and relative size are poor approximations, but relative size gets closer.
Large animals aren't just scaled up versions of smaller animals. Large animals use the space for more complicated digestive tracts. Their larger appendices allow them to have more joints with more degrees of freedom, and longer appendices and higher speeds require a higher resolution from their eyes to perceive sufficient detail. Skin with more surface area also means many more cells for sensing heat, pressure, damage, etc across the entire skin.
All of this means the brain of a larger creature needs more "IO ports", more neurons to compute sensible outputs for them even for routine tasks, and more neurons to break the wealth of data into a form usable by a cognitive process or an instincutal reaction (mapping to area of body, computing averages over time and over multiple inputs, correlating different types of senses, etc).
All of these differences have different scaling factors. The amount of neurons to count to 5 stays the same, the amount of neurons to process skin sensations scales approximately with skin area, or the square of animal diameter, but the amount of neurons to control motor functions or the digestive tract can grow much faster. Overall, putting brain size in relation to animal size or mass is a decent first approximation and works in practice.
Because not all of the brain is 'compute'. Lots of it seens to be allocated to map 1:1 with body areas (or probably the number of nerves in those body areas). I guess you could see this as IO.
probably because relative size is a good proxy for evolutionary "resources" being allocated to brains.
why absolute size doesn't matter as much as we think it should is a really good question, but clearly our modern transition from "the nervous system is like a steam engine" to "the brain is like a computer" is only an improvement and not actually a good model yet.
So, when "one thinks of compute capacity" one arrives at the wrong answer instead of one that fits the evidence and statements of the evidence seem like "a strange argument".
Large animals aren't just scaled up versions of smaller animals. Large animals use the space for more complicated digestive tracts. Their larger appendices allow them to have more joints with more degrees of freedom, and longer appendices and higher speeds require a higher resolution from their eyes to perceive sufficient detail. Skin with more surface area also means many more cells for sensing heat, pressure, damage, etc across the entire skin.
All of this means the brain of a larger creature needs more "IO ports", more neurons to compute sensible outputs for them even for routine tasks, and more neurons to break the wealth of data into a form usable by a cognitive process or an instincutal reaction (mapping to area of body, computing averages over time and over multiple inputs, correlating different types of senses, etc).
All of these differences have different scaling factors. The amount of neurons to count to 5 stays the same, the amount of neurons to process skin sensations scales approximately with skin area, or the square of animal diameter, but the amount of neurons to control motor functions or the digestive tract can grow much faster. Overall, putting brain size in relation to animal size or mass is a decent first approximation and works in practice.