Average number of neurons in the human brain = 100 000 million
For an informed technical discussion of the mammalian brain and its structures, see Finlay, Darlington, & Nicastro (2001). See also Klinowska (1994).
I am heavily indebted to Kinser (2000) for the following summary of attempts to correlate "intelligence" (or behavioural complexity) with different measures of the brain.
It has to be remembered that some animals have bigger brains, simply because they have bigger bodies. One would expect C. elegans (whose body length is less than 1 mm) to have a small brain. A honeybee's brain is small, but its weight is 1% of its overall body weight (cf. 2% for human beings). Whales and elephants have much bigger brains than human beings, but their bodies are also much larger. So, it would seem more logical to compare brain weight, as a proportion of total body weight.
However, many small mammals and birds have larger brains in proportion to their body size than human beings. Does this mean that they are more intelligent? It turns out that brain weight in vertebrates does not increase linearly with body weight, but exponentially. The following equation was developed in the late 19th century by the German physician Otto Snell:
E = C x (S to the power of r),
where E is the weight of the brain, S the body weight, C is a constant "cephalisation factor", and r an empirically determined exponential constant, whose value is about 0.66 for mammals. The value of the cephalisation factor is 0.07 for mammals and 0.007 for "lower vertebrates". However, certain species of mammals have brains that are larger than predicted for their body size, using Snell's equation: for instance, humans, dolphins, chimpanzees, rhesus monkeys, elephants, whales and dogs have brains that are 7.44, 5.31, 2.49, 2.09, 1.87, 1.76 and 1.17 times bigger than predicted for their respective body sizes. These animals have a high encephalization quotient (EQ). However, there is no evidence that mammals with low "encephalization quotients" are less intelligent than other mammals.
Cortical folding has been suggested as an indicator of intelligence, but studies indicate that cortical folding is mostly a matter of brain volume. Some mammals may show more or less folding than human beings, but there is no evidence that these convolutions correlate with a higher degree of behavioral complexity.
The neocortex is the structure in the brain that distinguishes mammals from other vertebrates, and it is widely assumed to be responsible for the development of intelligence. However, research has shown that the volume of a mammal's neocortex correlates roughly with the size of its brain. There is nothing exceptional about the size of the human neocortex; it is roughly what one would expect for a primate with our brain size (but not body size). Moreover, the differences in the size of the neocorticex (relative to brain size) between primates and other mammals are relatively slight.
Nerve cells have also been posited as the source of our intelligence. In the mammalian cortex, the number of neurons per unit volume (i.e. neuron density) actually declines with increases in brain size, but this is compensated for by an increase in neural connectivity. Overall, activity per unit volume of brain is independent of a species' brain size.
Are there, then, some better ways of measuring the brain's capacity to generate complex behaviour in animals?
Drawing on the experience gained from a lifetime of research, Bullock (2003) has summarised the results of comparative studies of the brain and nervous system in various taxa of animals:
It doesn't look as though the great leaps in brain develoment are due to sheer numbers or to more kinds of transmitters or modulators... What else could it be?
I made a stab some time ago to estimate the numbers of kinds of neurons, not just on morphology, cytology, and kinds of processes or on transmitters, modulators, chemical öreceptors, cytochemistry, molecular inhabitants, immunological criteria and so on - but on all of these and their permutations. Very importantly, one must add their dynamic personality traits - tendency to spontaneity, to bursting, to fast or slow adaptation or both in sequence, to facilitation or its opposite, to regular or irregular firing and each of some forty-odd additional integrative dimensions that show a distribution from high to low, characteristic for types of neurons....
Lastly, one must add the number of significantly different receptive fields and projection fields. Neighboring neurons have some degree of overlap but also some degree of non-overlap that makes it possible to recognize distinct sets of fully equivalent neurons from other such sets... When we speak of the numbers of kinds of neurons, we should mean on all these criteria.
My estimates, based on a few well studied systems, such as the visual system and a lot of educated extrapolation, put the number in the hundreds in some simple worms, thousands in Aplysia, tens of thousands in more advanced insects and crustaceans and in a fish like a carp, millions in a rat and billions in humans! ... This may seem out of reason and is certainly at odds with a common view, such as that of Changeux ... who said the numbers of categories of pyramidal cells in the cortex is probably tens or hundreds. He neglected to consider their connectivity at the level of receptive fields and projection fields and by "categories" meant classes. I am asking how many species are there, on all criteria, including distinguishably different connectivity fields.
I want to conclude by underlining the extent of our ignorance about the evolution of the nervous systems of animals, most especially in the intermediate integrative levels, the mesoscopic levels of small or larger assemblages of cells in systems between the subcellular, ionic channel and molecular level and the behavioral level. The tremendous gaps between the complexity of the brain in a lower invertebrate and a higher invertebrate or a lower vertebrate and a higher vertebrate are apparently not principally attributable to sheer numbers of cells or to properties of neurons, synapses, transmitters and modulators but to some organizational features that utilize the permutations of scores of integrative variables and thousands or millions of connectivity variables.
We are seriously lacking in knowledge of what the actual differences are between less complex and more complex brains, particularly in their physiology.
Number of neurons in a mouse brain = 1000 million
Number of neurons in octopus brain = 300 million
Number of neurons in honeybee (Apis mellifera) brain = 960,000
Number of neurons in fruit fly (Drosophila melanogaster) brain = 240,000
Number of neurons in cockroach nervous system = 100,000
Number of neurons in sea slug (Aplysia californica) nervous system = 18,000 to 20,000
Number of neurons in each segmental ganglion in the leech (Hirudo medicinalis, an annelid worm) = 350
Number of neurons in C. elegans nervous system = 302
(Sources: Chudler, 2003; Fell, 2001; Menzel and Giurfa, 2001; McGovern Institute for Brain Research at MIT, 2003.)
I'm talking about evolution of nervous systems, from simple to complex and looking for relevant measurable variables. Relative brain size or number of cells, or of synapses or of impulses per second in the whole system are doubtless relevant but imperfect and inadequate measures of the machine, ... just as numbers of distinct behaviors in the whole ethogram, including brain states such as stages of sleep, arousal and attention can represent an approximation to a measure of accomplishment of the machine - complexity of behavior. But both neglect major variables simply because they are difficult to count.