Does the brain really function like a computer as in, ultimately every response is related to a binary sequence based on whether particular neurons fire or not?
First of all, I would like to point out that making analogy between digital computers and the brain is often very misleading.
That being said, my answer is, some scientists believe so, some don't.
Several things to consider:
Some neural systems are not spiking. C. elegans for example has a nervous system that is entirely analogue. Human nervous system also contains neurons with graded responses (mostly in the sensory front-end though).
Spiking neurons may be binary at each time point, but time itself is continuous. Firing at 0.003 seconds later can represent something different. (in contrast to the usual synchronous digital architecture of computers)
The neuron doctrine is sometimes challenged. Glial cells that do not fire may have important functions for information processing. See:
- Bullock, T. H., Bennett, M. V. L., Johnston, D., Josephson, R., Marder, E., and Fields, R. D. (2005). The neuron doctrine, redux. Science, 310(5749):791-793.
While action potentials are usually binary, you should note that synaptic communication between neurons is generally not binary. Most synapses work by neurotransmittors, and this is a chemically mediated graded response that, for example, act on voltage-gated ion channels. So even though action potentials are often binary, communication between neurons are most often not, and action potential firing can involve the integration of synaptic information from many different neurons. Therefore, the brain as a whole cannot be reduced to a binary system.
See this as a complement to @Memmings answer.
John vonNeumann, the famous computer scientist, tackled this idea in his last book, 'The Computer and the Brain.' He personally landed on the side of the brain being a binary system, due to the behavior of neurons which either fire or do not fire.
While that is an important observation, and will have significant consequences to people trying to create artificial brains within computer systems, I think a more important observation has to do with computational complexity. It is very easy to build systems which, at least theoretically, have the potential to be universal computers. From that fact, it is fairly trivial to see that whatever definitions you choose to work with in terms of brain input and output (sensory nerve cells feeding electrical impulses from organs of perception being a possible definition of 'input' and propagated impulses to muscles, or changes in the neural structure itself being possible definitions of 'output' for example), yes it is possible to construct a binary system which can perform the same calculations as a human brain.
However, there is a catch. Because it is impossible to perfectly know the complete state of the brain, and because any degree of inaccuracy in the starting state of the binary system, no matter how small, will cause the behavior of the binary system to diverge completely from the behavior of the specific brain being modelled, it is reasonable to say that no particular individual brain can be reduced to a binary system.
otakucode Chapeau! This is a very deep and thorough answer, wich takes well into account the side of computational complexity (and computing machines theory). From a theoretic point of view, the Systems Theory shows that, no matter the complexity of a system, and no matter if the system itself is or is not digital (based on binary logic or analogic), it is always possible to translate it into a digital system, with a defined level of accuracy (overall quantization noise). The big issue in real world is to define and measure the state vector of a real brain, and even more difficult to define the component values of the state vector at time 0 (the concept itself of initial time is hard to define in this case). Under this point of view, the details of the biological structures and of the neural processes running in the bain, do not change the substance of the problem, only its complexity. To solve somehow the big issue of defining the state vector of the brain one day, would mean to have defined an acceptable logical schema of the wole brain. To approach that outcome a way bigger effort should be made in the direction of cathegorizing the different structures and processes of the brain into the blocks of a logical system comparable to a computer program, or if we prefer, to a finite state probabilistic digital machine, in which each block should embody a single structure or process of the brain. Working block by block and then combining and connecting them into the global schema should be the very long path to treck.
as far as I know the brain processes data in stages the neurons themselves are not purely binary as in a computer in that every action has a predetermined output. the neuron response tends to be goverened by sigmoid function output and hence the use of this function in artificial neural networks. further, the synapses have strengths that is depended on the amount of neurotransmitter in there which obviously varies from cell to cell and even in the same cell and hence one speaks of the probability of a neuron firing given a certain stimulus. additionally the neurons from sensory organs fire pulses at a frequency that increases with the strength of the stimulus. furthermore, the data from sensors is processed in layers of neurons lower layers have rapidly firing neurons whule higher layers fire at much lower rates.
you also have to consider the fact that the brain is actually a complicated network of "recurrent" neurons meaning that the output is fedback as an input and this is different from simple computer gates such as AND gates or XOR gates it is similar to counters may be but obviously on a very bigger scale. one more point is that recurrent neural networks have built in memory that enables the pattern recall and recognition and so the study of the brain as a binary system is very complicated and will fail to explain how the brain works.
on the macro scale the human brain operates as a bayesian inference engine more or less I mean when it comes to thinking and inference i.e. it relies on probabilities and knowledge gained from past experiences to deal with current problems and new data