Do computers use logic?

Question

I know we refer to computers as using logic, logic gates and the like, but is this just us ascribing human capacities to the machines? It sounds like a case of us giving more meaning to the machines than they deserve. I've read about things like derived intentionality and this seems to echo that.

Answer 1

Allow me to be precise about this. Logic (in the formal sense) is a system of manipulating symbols according to rules. Computers can manipulate symbols according to rules — that is more or less exactly what they are designed to do — so in that sense computers can use logic.

That being said, computers cannot (currently) reason. Reason entails more than the application of logic: e.g. valuation, goal setting, reflexive analysis, reversibility, error awareness, etc. A computer can manipulate symbols to get from A to B easily enough, but it cannot (currently) distinguish A or B from each other or any other proposition, cannot evaluate the importance or value of these proposition, cannot decide to do that manipulation on its own... You get the picture.

There are computers that are programmed to do mathematical proofs, and they occasionally create proofs that humans have been unable to solve. But they don't do it with intelligence; they do it with brute force. As a rule mathematicians dislike computer-generated proofs; not out of some prejudice, but because CG proofs are ugly and inelegant, the kind of disorderly mess we get when one simply hammers one's way through a problem. When (and if) computers are capable of choosing beautiful, elegant, and efficient proofs over ugly, inelegant, sprawling proofs, then we might start talking about computers using reason.

Answer 2

You might like this answer: Why is a measured true value “TRUE”?

Physicist Richard Feynman makes the case in this lecture that computers are essentially limited to sorting: Hardware Software & Heurustics (from 1986).

logike, "branch of philosophy that treats of forms of thinking; the science of distinction of true from false reasoning," from Old French logique (13c.), from Latin (ars) logica "logic," from Greek (he) logike (techne) "(the) reasoning (art)," from fem. of logikos "pertaining to speaking or reasoning" (also "of or pertaining to speech"), from logos "reason, idea, word"

From Etymonline

Do computers use 'forms of thought'? You can split hairs semantically if you like. But I suggest they are able to make inferences by applying rules in exactly the same way humans do using formal logic.

The Chinese Room Argument applies, and the syntax semantics distinction. But I'd say 'form of thought' is syntax.

Answer 3

Computers use logic, but differently than people and other Great Apes.

Logicians make a distinction between formal logic and informal logic. Computers are fully capable of doing formal logic, but they struggle with informal logic because while formal logic is a syntactic system which is easily done by substituting symbols for other symbols, informal logic relies heavily on domain-specific semantics, including what some call the material logic of natural language. In the latter category, it's possible to build sophisticated software to replicate informal logic, but when done, it requires a lot of work from people to help codify extra rules, values, and language to emulate informal logic that human genes do with ease for human beings. Expert systems are a special sort of AI that weighs through facts and draws inferences, and such systems ultimately rely on transistors and machine code. Of course, the human brain ultimately relies on neurons and epigenetic components of the nervous system, so human beings also have a physical basis.

Now, 50 years ago, there was a clear line between human reasoning and machine reasoning because the rules that were implemented had to ultimately be crafted by human programmers. However, today, the game has changed somewhat with connectionist models asserting some influence in systems. For instance, thanks to the ever-widening collection of machine learning (ML) techniques, there are now rule-based machine learning models:

While rule-based machine learning is conceptually a type of rule-based system, it is distinct from traditional rule-based systems, which are often hand-crafted, and other rule-based decision makers. This is because rule-based machine learning applies some form of learning algorithm to automatically identify useful rules, rather than a human needing to apply prior domain knowledge to manually construct rules and curate a rule set.

What does this mean? It means machines are learning in the same way people do, that is through induction by repeated observations using information from the outside world. A computer vision system which learns faces does manifest intentionality in the same way a dog, chimp, or child manifests intentionality.

So, machines, particularly with systems such as cameras, sensors, and other robotic peripherals can learn from the environment, devise their own rules, and then use those rules to make decisions. That's the essence of how and why human's use logic. While we get our physical endowment from genes which have evolved, our machines are still largely designed, though there is also an increasing repertory of computational strategies to evolve design such as the use of genetic algorithms and ML strategies that mimic natural processes. While clearly the how's and why's are different from human beings and our use of logic, there is no disputing that sophisticated robotic systems that are designed to mimic human reasoning can and do use logic. "Using logic" is not a yes-no question, but one of degree, and there's little sign that computers and robots won't continue to use logic more and more like people as the systems become more sophisticated and research continues forward into understanding how the human brain and its neurons compute to provide human-level intelligence.

Answer 4

From Wiktionary:

Noun
logic (countable and uncountable, plural logics)

1. (uncountable) A method of human thought that involves thinking in a linear, step-by-step manner about how a problem can be solved. Logic is the basis of many principles including the scientific method.

2. (philosophy, logic) The study of the principles and criteria of valid inference and demonstration.

   2001, Mark Sainsbury, Logical Forms - An Introduction to Philosophical Logic, Second Edition, Blackwell Publishing, page 9:

   "An old tradition has it that there are two branches of logic: deductive logic and inductive logic. More recently, the differences between these disciplines have become so marked that most people nowadays use "logic" to mean deductive logic, reserving terms like "confirmation theory" for at least some of what used to be called inductive logic. I shall follow the more recent practice, and shall construe "philosophy of logic" as "philosophy of deductive logic".

3. (uncountable, mathematics) The mathematical study of relationships between rigorously defined concepts and of mathematical proof of statements.

4. (countable, mathematics) A formal or informal language together with a deductive system or a model-theoretic semantics.

5. (uncountable) Any system of thought, whether rigorous and productive or not, especially one associated with a particular person.

   "It's hard to work out his system of logic."

6. (uncountable) The part of a system (usually electronic) that performs the boolean logic operations, short for logic gates or logic circuit.

   "Fred is designing the logic for the new controller."

It is senses 1, 2, and 5 in which we use the word "logic" when discussing human behaviour. Specifically, deductive logic and inductive logic, which fall under sense 2, are the kinds that philosophers talk about.

It is senses 3 and 4 in which mathematicians use the word, also called "axiomatic logic" or employment of an "axiomatic system". One such system is that of Boolean algebra. These usages of the word to refer to the application of axioms to derive other true statements within that axiomatic framework naturally arose from the earlier sense in which "logic" refers to deductive reasoning.

It is senses 4 and 6 in which computer scientists use the word, and sense 6 is almost exclusively the sense in which computer hardware / embedded systems engineers and computer programmers use it. Specifically, the hardware of modern computers implements the logic of Boolean algebra. That is the job of so-called "logic gates".

See also: Rewriting and abstract rewriting systems

Answer 5

Computers do not use logic. They do however, use logic GATES. Logic gates perform very small pieces of logic calculations like AND, OR and NOT, but this is not enough to say they use logic.

Computers follow instructions. It is possible to program a computer to perform pretty advanced logic deductions, but most computers don't do this. You could say that these computers use logic, but they are an exception, not the rule.

Answer 6

Difficult. I mean the question is basically where the "mind" or "soul" of the "entity" of a computer is located or whether it has or is any of that to begin with.

Like if you take your device and consider it a blackbox then somewhere inside of "it" something is happening that can be described by logical operations, so "it" "does" apply logical principles somewhere (by choice or accident).

The problem is that when you take away the blackbox and look at the inner workings then there is no agent and none of the components immediately presents itself as having the capability of being an autonomous agent. So there is nothing that could actively "use" something.

And if you think that electricity is some quantum black magic hocus pocus and that there's some hidden nonsense going on... Well you could build mechanical computers, water based computers and it hardly gets more transparent than this plastic based one running the game "nim".

So let's consider this case of Dr. Nim. Like where is the logic happening. Is it the user? The user sets things in motion with their input but they aren't doing any of the computation or apply logic. In fact the user just needs to be able to push a button and flip a switch no deep thought or logic required. Is it the current? In this case a marble? A MARBLE? While it moves the pieces it certainly isn't the marble acting logically. Is it the plastic thing itself? Kinda... sorta... kinda not.

Like it's something that moves and it moves according to logical rules, but actually it only moves by mechanical rules that aren't really "followed" by an agent, but just confirm to the laws of physics. If you'd turn it upside down or employ it in space it wouldn't work. That being said if you'd put a human into a hostile environment they also might seize to work.

To a degree the logical work is done by the programmer/engineer who constructed the gizmo, nevertheless when it's actually used they are no longer present. So for this creation, their "creator god" may as well be dead and it would not interfere with the scenario, so how important can they be? The logic is contained in the device itself not merely in the creator of that device.

Now for the sake of argument you could abstract from the device the algorithm that is at the heart of it and consider that to be the computer, in which case the user would be the one applying logic by following the logical steps while the program itself would not be using logic it would be an implementation of logic. However as we've already covered the user could as well be incapable of logic and is just starting the process while the device performs the heavy lifting. But is it "USING" logic? "IS" "IT", to begin with?

Now so far we could put the blame on the programmer and argue the computer is just a manifestation of an algorithm and thus is usage of logic but doesn't use logic.

However that doesn't quite cut it either anymore, because what about machine learning and self-programming computers? Like we can construct systems that improve their mechanism on the fly by being given a desired output, a capability to produce outputs, a set of parameters and a comparison function. Then an output is produced, it's compared to the desired output, if better suited, keep it if unsuited discard it and try again with different parameters, if close enough stop.

Now the programming part is no longer the part of the programmer. In fact it's possible that the program solves problems that were too complex for the programmer themselves. At the same time it's still the same non-thinking machine. The game might have become more complex but the concept remains the same. It's still following an algorithm, still only an application of logic in that maybe not even logic but probability. That being said when pushed to the extreme it could be chance find logical properties, like if you want it to present you with a decision tree, then it's perfectly possible that what you end up with is a logical diagram. So it's not just an implementation of logic but it creates implementations of logic.

Though given that it's still just passive material, does that count as "using" logic? And if it doesn't does our application of algorithms count as logic? Because in the end our "usage" mostly rest upon our biased perception of ourselves as agents rather than passive material, but if passive material is capable of such feats where is the line between the two?

Also even if there is a line and we are active and computers are certainly not, given that they produce logic and can be understood with logic, is it useful to think of them in terms of entities using logic? Like is our process of creating logical constructs similar enough for that analogy to be useful regardless of whether it holds.

And here it might come back to the programmer/engineer, because the computer way of thinking is largely modeled after our own way of thinking, though much of the programming currently happens at a level that is far removed from the actual processes so while it might have had been useful, we might move into a direction where it no longer isn't or we might coincidentally move in the same direction as our own way of thinking. But unfortunately as of right now we don't seem to know how our own mind and consciousness works to begin with so that a comparison to our most capable tools is always attempted but it's never clear as to whether it actually works like that.

Answer 7

I think it is appropriate to quote George Boole here:

No general method for the solution of questions in the theory of probabilities can be established which does not explicitly recognise, not only the special numerical bases of the science, but also those universal laws of thought which are the basis of all reasoning, and which, whatever they may be as to their essence, are at least mathematical as to their form.

The way I understand this is that he is proposing there are "universal laws of thought", of which the human mind as well as a computer are physical implementations. This is meant with regard to their ability to work with logically provable statements, not creativity or the ability of abstraction, which classical computers are completely lacking. Any problem a computer should solve has to be "presented and tailored to it" specifically by humans, but that does not take away from their ability to actually solve "logical problems" according to "laws of thought".

Answer 8

Computers intensely use formal logic, which is built into them. Formal logic amounts to elementary arithmetic (AKA boolean) with the binary digits 0, 1, and the operators not, and, or.

Combining these operations in various ways, you reconstitute all arithmetic on integers and other numerical representations. All that computers can do is summarized as handling

• numbers (numerical applications),
• characters (text processing),
• programs (textual sequences of instructions to be executed automatically).

Programs do use a more advanced form of logic, closer to what we call first order logic, and programming is a task similar to theorem proving, which is still completely formal. But programming is anyway performed by humans, not by computers.

So, no, computers do not use logic in the sense of a cognitive activity: computers do not think, they have no mind, no consciousness, no nothing. But due to their speed and immense data storage capabilities, they achieve outstanding tasks that a human could never perform, just using zeroes and ones.

---

The question can be re-examined in the light of the famous Artificial Intelligence techniques, which aim at handling "knowledge" rather than "data". But so far, not a single droplet of intelligence has been obtained.

Answer 9

Most answers are confused, and the problem is simple.

• Logic is the FORMAL set of rules that govern reason.
• Reason is the potential to think.

I know we refer to computers as using logic, logic gates and the like, but is this just us ascribing human capacities to the machines?

No. Machines implement rules. We would be "ascribing human capacities to the machines" if we would say machines "think" or "reason". But they DO implement and apply rules, which imply they apply Logic.

It sounds like a case of us giving more meaning to the machines than they deserve.

Badly expressed, but your point is somehow understandable. If machines implement Logic, and we want to communicate about it, we just do. That is, machines "deserve" it.

Answer 10

@Ted Wrigley 's answer is excellent but it's omitting another hidden sub-question inside the original question: Neither the question nor the answer differentiate between what "bare computers" can do and what "programmed computers" can do.

A programmed computer can do anything. It can do all of the "cannots" that have been listed (that is, with enough programming efforts). When does it stop being "just logic" when a sufficient amount of logic gets used to create "reason"? You decide.

However, we might assume that the original question was asking only about what "bare" computers can do. It could be rephrased as "is computers' core design somewhat resembling logic?".

Under that assumption, everything Ted said is valid : Yes, they are built around logic principles, but no, "reason" is not part of the core design.

Answer 11

Digital Logic

https://cs.lmu.edu/~ray/notes/digitallogic/

Believe it or not, we can model all computational processes (that we know of) by operations from, of all things, plain-old classical logic.

The reference shows how digital gates and other devices in digital computers are used to implement bivalent logic. These devices do not manipulate symbols according to rules because a symbol is an abstract sign recognized by a human intelligence (HI) or general artificial intelligence (GAI) if such entity ever emerges. The devices implement logic functions using physical system states and deterministic state transitions. We map symbols to the physical system states to program the logic gates or the computer which then performs rapid logic functions.

Logical Neural Networks

https://skirmilitor.medium.com/logical-neural-networks-31498d1aa9be

Our first main idea is the use of constrained optimization during learning to guarantee logical behavior of all neurons in an LNN.

Due to their 1-to-1 correspondence with systems of logical formulae, an LNN can also be viewed precisely as a collection of logical statements in real-valued logic (where truth values are not restricted to just 0 and 1 but can lie anywhere in between). In a separate, very fundamental work [11] we have shown that the larger class of logics to which LNN belongs is capable of sound and complete (i.e. correct and thorough) reasoning, to the same degree as has been shown for classical logic [ref: real-valued logic foundations blog].

Thus, like the famous ‘wave-particle duality’ of physics, LNNs can simultaneously be seen entirely as neural nets and entirely as sets of logical statements, and thus able to leverage the capabilities of both the statistical AI and symbolic AI worlds.

Before we can discuss LNNs further, it is necessary to develop an understanding of the computations their neurons perform. The output of a logical neuron is computed much the same as it is for any neuron by applying some nonlinear activation function f : ℝ → [0, 1] to a linear combination of its inputs w · x − θ for an input vector x, weight vector w, and activation threshold θ, as shown in Figure 2. Different from other neural nets, however, is that LNN’s neural parameters are constrained such that neurons behave according to their corresponding logical gates’ truth functions.

This reference refers to the effort to implement human logic and reasoning via so-called Logical Neural Networks. When humans interact with digital computers we can decode the explicit programming code or digital circuit logic functions. If LNNs begin to perform block functions of logic or reason in a limited domain then we have a black box that implements a function of logic or reason.

Complexity

There are problems that define system states and deterministic changes of state, but which are sensitive to both initial conditions and to unavoidable computation error associated with the use of rounding and truncation methods. In these cases if we run the automated computer algorithm, using the same initial conditions, to solve the problem for the path of evolution of the system, then we get different paths each time we run the computation. This gives rise to concepts of chaos and complexity theory in the domain of computational theory.

Answer 12

Monkey-Man tears off an apple by own forces.

Man-Monkey takes a stick to tear off an apple by own forces.

Man create Logic.

Man took a stick to force a human to tear off an apple by his forces.

Man call this stick - Boomstick.

Man create a logic rules to force the robot-computer to took a stick to tear off an apple.

But robot computer is too expensive, then step back.

Man create a logic rules to manage a computer that force a human by logic rules, (that a human accepts) to took a stick to tear off an apple by his forces.

Does computer understand man's Logic?

No, but it use it's rules, same as human.

What ll Man do if logic rules will not works on humans??

Oh, Man still have Boomstick.

L=logic...

"a bit of the old ultra-violence" no one cancels.