I'm wondering about the fundamental differences between human thinking and computer thinking. Is it wise to consider all human thinking in terms of algorithms?


This line of thought has a rich tradition in philosophy. One potential way to gain empirical evidence one way or another is through the famous Turing Test, which challenges a computer to successfully imitate the perceptible output of human thought processes effectively enough that a human being would be unable to tell the difference.

To the extent that a computer can imitate the output of human thought processes through algorithmic means, it is suggestive that human thought is likewise algorithmic.

It's worth noting that this test can only establish a positive result (there aren't any conditions established for decisive failure). It's further worth noting that even the best candidates so far have fallen far short of the standard. This may reflect the limited complexity of the computer versus the brain, but it might also indicate that human thinking is in fact non-algorithmic.

In the larger philosophical landscape, both Turing's original test and this adaptation of it are squarely in the British Empiricist tradition of insisting that the only meaningful statements about internal processes such as thought are the ones based on the empirical effects that can be perceived and measured externally.

edited to respond to concerns by @adeena

  • 1
    Right... and a failure to fool the judge does not mean that the computer is not intelligent. :) BTW... while I love to debate Turing, I'm not sure it's helpful to the original question. I really think the book "Intelligence" by Jeff Hawkins is a very worthy read for anyone interested in understanding human intelligence and how the brain works (whether or not it can be considered algorithmic, etc)
    – adeena
    Feb 13 '14 at 18:58
  • 1
    Interesting take given that in the into of Turing's paper, he starts out posing the question: "Can machines think?" (not "Do humans think algorithmically?" or anything like that...) :)
    – adeena
    Feb 13 '14 at 19:16
  • 1
    @adeena Upon further research, you are correct. I will withdraw my earlier comment and further edit my answer to reflect this. Feb 13 '14 at 20:16

No, I don't think it is wise to see emotions and intelligence as algorithms. It certainly isn't the traditional way to think about the world of human experience.

One reason to think of algorithims is that one can do things with them. Crucially this aspect is missing from intelligence seen as algorithmic.

One can certainly simulate certain kinds of thinking with algorithms but this does not show that they are the same thing. What one is doing is modelling. Physics, for example models the physical world - but one should not then make the mistake that the physical world is physics.

The Turing test takes a functionalist view of intelligence. Quite crucially Turing ignores the fact of inner life. I'd argue that thinking is intrinsically tied up with inner life. To ignore it doesn't mean that its not there, or that its not important, but renders the question more tractable.

  • "One can certainly simulate certain kinds of thinking with algorithms but this does not show that they are the same thing." - what about arithmetic? Can a computer perform arithmetic, or just simulate it by processing symbols?
    – obelia
    Feb 17 '14 at 2:20
  • Good question. I'd say not. Because you & I know that its doing arithmetic. But the computer doesn't. Its simulating it. Without the semantics given to it by us - its merely outputting arbitrary symbols on screen. It seems counter-intuitive because it looks like it is exactly doing that. Feb 17 '14 at 3:14

1) I highly recommend reading the book "Intelligence" by Jeff Hawkins

2) As for the Turing Test, it said nothing about how the computer performed the actions and it had nothing to do with emotions. It was simply whether or not a judge could discern between a computer (a "machine") and a human based on answers to questions.

2a) In practice (...and I've participated in some modern version of a Turing Test called the Loebner Prize...), to win the Turing Test and attempt to fool the judge, all we wind up doing is trying to find ways to make the computer a better liar/trickster and we really aren't making any progress in artificial intelligence this way.


  • 4
    The philosophical point is that for Turing, fooling the judge IS artificial intelligence. One might certainly disagree with that claim, of course, but it's the philosophical underpinning of the original Turing Test concept. Feb 13 '14 at 18:48
  • I agree except for the part of your answer: "successfully simulate human thoughts, emotions, etc". None of that is required or necessary to fool the judge. :)
    – adeena
    Feb 13 '14 at 18:55

Saying 'computers can think', meaning: thinking is really just mechanical.

Or else, meaning: computers are becoming really sophisticated, having near human behaviour.

Neither of these things is really true. Thinking, thoughtfulness is by definition not mechanical. And no matter how sophisticated or advanced we consider computer behaviour, can you even count how many human behaviours a computer lacks?

Asking whether computers think is not an empirical question, but a question about what we will count as 'thinking'.


Human brain has evolved over several millions of years. Thus we have a very complicated thought process. Many thoughts have deeper meanings but few are trivial. Coming to computers their thought process is very young. Human thinking cannot follow a strict algorithm but a very loosely based model dependent on several parameters. It is more closely related to statistical probabilities or stochastic process.

Before I go deeper into the topic I would like to frame few postulates: 1) Survival is one of the integral part of human thinking. One will never think of jumping from a building(Exceptions are present but probability tends to zero).

2) Survival of oneself can be sacrificed when one's offspring is in danger and lot energy and resources are spent to raise them. Also given the fact that one cannot give birth to offspring. This theory is behind culture and ethics of human race. Raising a human baby is extremely difficult so culture and ethics had to be created to ensure the survival of the baby. Physical, mental and emotional support can be given this way.

3) Survival of oneself and one's offspring can be sacrificed when one poses a threat to the survival of one's community/species. This is associated to suicidal tendency. But again fact number 1 and 2 prop up. This leads to rebellion. To counter suicidal tendency, rationalizing and boasting is an effective short term tool. But sub conscious mind is still in suicidal mode. These leads to apathy and extreme violence.

After several billions of iterations of these events(evolution), human thinking has developed. Comparing this to computer programming is absurd. As the survival is never an integral part of computer's thinking. Not performing a task is not the end of the world for computer.

Turing test can be passed by a computer if it can show the respective states of insecurity. When one raises a voice we expect a reaction from other side. If an AI program is used we may not be able to simulate this reaction in the right way.


Write a program in machine or assembly language. You will immediately realize computers do not do anything that vaguely resembles thinking. Some software such as Watson can mimic human thought but it is without consciousness, the prerequisite for thinking. My dog and cat think. My computers never do. Can computers have knowledge? Human thinking is primarily concerned with knowledge, it acquisition and it's use. Questions & statements. Computer software can also feign knowledge. The computer does not know it is representing knowledge. The computer cannot be conscious of wholes which we use all of the time. Computers only process binary data. That is all

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.