What is the purpose of sentience?

[GladScientist]

What function does it serve? Assuming that my brain is just a computer run entirely by deterministic and/or probabilistic processes, why does there have to be a "passenger along for the ride"?

When I talk about sentience, I am referring to your ability to experience experiences. Many would say that it allows us to think or make decisions, but our computers do this just fine and they are not aware of themselves. Isn't the human body and brain just a machine that responds to input that it receives through the senses? Why does there have to be an entity within that experiences this life?

 

[atyy]

You may find relevant speculation in
Churchland, Self-representation in nervous systems
Holland & Goodman, Robots with internal models

Of course, http://condor.wesleyan.edu/courses/2004s/ees227/01/spandrels.html everything is an adaptation.

 

[Vendicar]

It is hard to know where your hands and feet are unless you are aware of where your hands and feet are.

Sentience derives from that requirement.

 

[GladScientist]

Vendicar, couldn't our body do all these things the same way that a computer does them? Computers are not conscious, and are capable of intelligence, and decision making. Why doesn't the brain just run on algorithms such as "BEAR = RUN"?

Or does a computer just develop sentience once it gets intelligent enough? Or are "senses" required to exist?

 

[atyy]

I think the question is very hard to answer, because we don't have an operational definition of sentience. Typically, to answer a question in science we have to be able to say if you do the following experiments and get these results, then by the data are consistent with a theory in which X is true. But we don't have experiments that are able to tell us whether a thermostat or a human being is sentient or not.

One note is that there are many "qualities" which are not either-or. For example, the distinction between liquid and gas. You may encounter regimes where it looks like the distinction is sharp. Yet if you go around the critical point, there will not be a phase boundary. And if you look closely enough, even the phase boundaries are not perfectly sharp in real life, since they require an infinite number of atoms, whereas real liquids only have ~10^23 atoms.

 

[Evo]

Computers are not conscious, and are capable of intelligence, and decision making.

Computers run programs and are limited to the information they are given. Also, you need to back up any extraordinary claims of fact with published mainstream scientific papers that back you up.

 

[Dr_Morbius]

You're assuming that animals don't have any consciousness. It's probable that animals have different levels of consciousness. Humans probably have the highest level of consciousness on Earth but other animals probably have consciousness appropriate to their lifestyle. I'm convinced that my cat, for example, does not just react to stimuli. She does thing that, to me at least, appear to involve decision making. In the same way that I might want to read a book or watch TV my cat does similar things in terms of her normal day to day routine.

 

[Vendicar]

Bear = run

What direction? Into a tree? Into a wall? Into the arms of another waiting bear?

The signal is bear. The correct response however is dominated by the noise of the environment around you.

Running away at 180' from the bear is probably the best choice until the bears become smart enough to hunt in packs and smart enough to know your programmed response to turn 180'.

They you are soon extinct as a species.

Programmed responses are what machines currently rely upon, and if you notice, you might see that they make pretty poor robots for that reason.

 

[Vendicar]

"Or are "senses" required to exist?"

The brain thinks by modeling the world. It does so through modeling the sensory inputs available to it. Without sensory inputs, you can not model the world in that manner and will have difficulty in thinking about the world around you since you will have not adequately experienced it.

Think about changing the inner tube on a bicycle. Your plan will involve a visual model of how to move the tools, how the tire will look when removed, how the tube needs to be placed, etc.

It will also involve a sense of where your hands are in space. How they are oriented to hold the tools, what pressure is used to perform the tasks etc.

You might also model the smell of the new tire, the sound of the spokes, or the sprocket retaining nuts entering a magnetic tray, or your pocket, or the ground.

You may also model getting oil on your fingers, and then on your shirt, and watch yourself place that shirt into the washing machine, or washing it by hand at a sink.

All of the planning for changing the tire is sensory modeling. You may throw in a word or two in your own inner dialog to keep track of things. But that too is a model of the sensory input of hearing.

Without such inputs, Computers will continue to have difficulty interacting in the real world.

 

[Vendicar]

All higher order animals are both conscious and self aware. That is self evident from their behaviors.

Dogs and cats suffer from depression, and experience love, joy and loss, bliss, anger, frustration, and all of the other emotions that people have.

These emotions exist for evolutionary purposes, and are probably heightened in social animals where they are "used" to provide cohesiveness to the group.

Where people differ from other animals is in the self recognition of those emotions and their subsequent heightening by means of self reinforcement due to that recognition.

Even crows have been shown to have an advanced sense of self awareness. Like Elephants they even recognize themselves - along with errors in their plumage - in mirrors.

 

[-Job-]

Bear = run

What direction? Into a tree? Into a wall? Into the arms of another waiting bear?

A modern phone could probably make that decision for you very satisfactorily, without the (apparent) need for sentience.

The OP asks a good question because AFAIK there's no established or even theoretical connection between sentience and computability or efficiency, which is at the heart of the problem.

 

[Vendicar]

"A modern phone could probably make that decision for you very satisfactorily"

Why would a modern phone want to?

 

[JorgeLobo]

"Purpose" is not appropriate unless you're buying into some superior being who/that rationally established the capability.
Reference to "function" is much better and you answer your own question in description.

 

[DaveC426913]

"A modern phone could probably make that decision for you very satisfactorily"

Why would a modern phone want to?

...kind of missing the point...

The point was that many kinds of decisions can be made with a very simple processor. The suggestion that animals need to decide to evade a threat does not lead to the need to sentience.

 

[Ryan_m_b]

There is some data suggesting that consciousness is a product of epiphenomenalism. In other words consciousness actually plays no part in the decision making process, rather conscious thought lags behind. However there remains an illusion of consciousness being the decision making agent. The data isn't solid yet and it's still largely an unexplored area but it could suggest that consciousness is a by product of intelligent agency rather than the agent itself.

 

[nucl34rgg]

What function does it serve? Assuming that my brain is just a computer run entirely by deterministic and/or probabilistic processes, why does there have to be a "passenger along for the ride"?

When I talk about sentience, I am referring to your ability to experience experiences. Many would say that it allows us to think or make decisions, but our computers do this just fine and they are not aware of themselves. Isn't the human body and brain just a machine that responds to input that it receives through the senses? Why does there have to be an entity within that experiences this life?

We have to be careful when we ask questions about the "purpose of something." "Why" questions are inherently unscientific. I could present an evolutionary argument for how sentience could prove to be advantageous to an organism's survival. The answer to these types of questions can normally be explained using evolution. Sentience served the function of helping your ancestors grow and reproduce more effectively in the environment in which they were in. Our computers do not think and make decisions "just fine." Sentience may very well be a key component of what we call intelligence.

 

[Vendicar]

"The point was that many kinds of decisions can be made with a very simple processor." - Dave

And they are generally made with a very simple processor - the spinal column.

But you miss the point entirely. Why should a non-sentient organism associate a bear with a complex pattern of response that is tailored to it's environment? And how would such a system evolve without higher level processing.

Why would a cell phone wish to escape from a bear? How would it evolve such a response without self awareness?

Is the cell phone going to fear being eaten? How and why? Fear itself is a manifestation of sentience.

 

[Ryan_m_b]

"The point was that many kinds of decisions can be made with a very simple processor." - Dave

And they are generally made with a very simple processor - the spinal column.

But you miss the point entirely. Why should a non-sentient organism associate a bear with a complex pattern of response that is tailored to it's environment? And how would such a system evolve without higher level processing.

Why would a cell phone wish to escape from a bear? How would it evolve such a response without self awareness?

Is the cell phone going to fear being eaten? How and why? Fear itself is a manifestation of sentience.

Again I think you are missing the point. The argument being presented to you is that an artificial intelligence can be programmed to react to a situation and potentially even be programmed to learn/adapt. It does not need emotion or instinct to make it act, action is simply a part of it's programming. And as objects like computers have no consciousness that we can identify the argument being put to you is suggesting that consciousness (and emotion) are not a necessary partner of intelligence.

Also please learn to use the quote button. If you want to respond to someone press the quote button on the bottom right of their post.

 

[Vendicar]

"..consciousness is a by product of intelligent agency rather than the agent itself." - Ryan

Clearly it is both an observer of decision and a decision maker. I have made a conscious decision to have eggs for dinner. That conscious decision will drive my response. At other times, when I am driving myself through the day, my consciousness monitors my automatic processes and on occasion modifies them for my conscious amusement or to further some plan of action.

Still the component actions themselves are unconsciously driven.

Now it could be argued that my decision to have eggs for dinner wasn't so much a conscious decision but just the monitoring of the result of a weighting of various factors by my unconscious. But if that is the argument (and I agree with it), I have to wonder if such a position pushes the definition of consciousness into a position where by definition the supposition that by definition no conscious decision is possible.

So what do you mean by conscious decision?

 

[Ryan_m_b]

"..consciousness is a by product of intelligent agency rather than the agent itself." - Ryan

Clearly it is both an observer of decision and a decision maker. I have made a conscious decision to have eggs for dinner. That conscious decision will drive my response. At other times, when I am driving myself through the day, my consciousness monitors my automatic processes and on occasion modifies them for my conscious amusement or to further some plan of action.

Still the component actions themselves are unconsciously driven.

Now it could be argued that my decision to have eggs for dinner wasn't so much a conscious decision but just the monitoring of the result of a weighting of various factors by my unconscious. But if that is the argument (and I agree with it), I have to wonder if such a position pushes the definition of consciousness into a position where by definition the supposition that by definition no conscious decision is possible.

So what do you mean by conscious decision?

If you read the link I provided epiphenomenalism suggests that conscious decisions are an illusion. It is proposed that all decision making is subconscious and the body moves to act on those decisions milliseconds before conscious thought of it occurs. This would suggest that consciousness is sort of like a real time memory rather than the decision making process itself. A good book that explores this is Blindsight by Peter Watts.

If you are having trouble getting your head around this think of it this way: there is a car with a driver. The driver uses their hands and feet to turn the wheel, press the peddles, change gears etc. What they don't realize is that the car is driving itself and all of the buttons, peddles, wheel etc are actually moving slightly before he touches them. But as he never stops trying to drive the car he never realizes.

Now this isn't fact, there is some suggestive data and the jury is still out.

 

[Vendicar]

"The argument being presented to you is that an artificial intelligence can be programmed to react to a situation and potentially even be programmed to learn/adapt." - Ryan

If it is programmed to have a specific response, then in what way is it intelligent?

You are also confusing intelligence with sentience.

 

[Ryan_m_b]

"The argument being presented to you is that an artificial intelligence can be programmed to react to a situation and potentially even be programmed to learn/adapt." - Ryan

If it is programmed to have a specific response, then in what way is it intelligent?

You are also confusing intelligence with sentience.

No I am not. What I suggested was if the programming allowed it to judge the situation, choose the optimum path to achieve the set goal (perhaps even deciding its own goal in a similar process), execute a set of behaviours and then analyse what happened to refine the decision making process in the future. These are the actions of intelligent agents and whilst we have never made sophisticated agents of the nature we are talking the point is there is no indication that conscious perception is necessary for intelligence. This is partially linked to the hard problem of consciousness, considering we don't know how consciousness is generated it is difficult to talk of an intelligent agent en par with a human (whatever that means) and whether or not it would be conscious.

Again, please use the quote button from now on. You can see the button on the bottom right ------------------------------------\/

 

[Vendicar]

It does not need emotion or instinct to make it act, action is simply a part of it's programming.

Without a sense of self awareness, how will it know that it's legs aren't already in the bear's mouth, and that as a consequence running is no longer an option?

Without a sense of self awareness, how will it know that it has hit a tree and can no longer advance?

Without a sense of self awareness, how will it know that it is far enough away from the bear to stop escaping?

Without a sense of self awareness, how will it know that it has broken a finger while climbing a tree to escape from the bear, and therefore should not attempt to use that finger to support itself?

Without a sense of self awareness, will it learn that being eaten by a bear is a bad thing? Maybe it is an essential roll in the cell phone's reproductive cycle.

Trees are not self aware. Hence they don't run from bears.

 

[Vendicar]

What I suggested was if the programming allowed it to judge the situation, choose the optimum path to achieve the set goal (perhaps even deciding its own goal\/

Why would the cell phone have a goal of not being eaten by a bear? How would the cell phone not know that the bear was a pet, or a vegatarian, or an antidisestablismenphoneatarian?

While you might get a program to emulate the desired behaviour you would spend the rest of your life writing it, and it still wouldn't cover all the desired behaviours unless you program self awareness.

 

[Ryan_m_b]

Now you are conflating awareness with consciousness.

Without a sense of self awareness, how will it know that it's legs aren't already in the bear's mouth, and that as a consequence running is no longer an option?

Presuming this agent has some faculties of sensory perception, has a memory and the ability to draw from this memory, model the world around it, model possible scenarios based on previous knowledge and execute behaviours towards an end goal why is consciousness needed?

Without a sense of self awareness, how will it know that it has hit a tree and can no longer advance?

If it has a sense of perception then it will know it is not advancing. If it has a memory, the ability to model the world and the ability to learn from memory then it can learn to avoid things.

Without a sense of self awareness, how will it know that it is far enough away from the bear to stop escaping?

How does awareness help with this exactly? It will know if it is far enough if it has both the ability to accurately perceive and model the world and can draw from experience/knowledge. Still not seeing why consciousness is needed.

Without a sense of self awareness, how will it know that it has broken a finger while climbing a tree to escape from the bear, and therefore should not attempt to use that finger to support itself?

Again; if the agent has perception, the ability to model the world and can draw from experience and knowledge (presuming it has the required experience and knowledge) then it can adjust its plan on the basis of the new conditions. Still not seeing why consciousness is needed.

Without a sense of self awareness, will it learn that being eaten by a bear is a bad thing? Maybe it is an essential roll in the cell phone's reproductive cycle.

It depends on what its values are. If we program said agent with the value to live and the capability to perceive, model the world, draw from memories of experience/knowledge and execute plans of behaviour then I see no reason why this agent shouldn't be able to learn how to avoid a bear. Still not seeing why consciousness is needed.

Trees are not self aware. Hence they don't run from bears.

Trees do not run from bears because the evolutionary history of plants has not included the evolution of a nervous system to be able to model the world etc.

You are conflating lots of things here. The conversations that have gone on in this thread are trying to explore whether or not consciousness is necessary for an intelligent agent or if it is just a factor of our evolution. I would advise you read some of the links I have provided (the blue, underlined words/sentences in my previous posts) so that you can learn some of the things that I and others are trying to point out.

 

[Vendicar]

If it has a sense of perception then it will know it is not advancing. If it has a memory, the ability to model the world and the ability to learn from memory then it can learn to avoid things.

Since it is a part of the world, and it would have the ability to model the world, it will necessarily need to model itself and hence it would necessarily be self aware and sentient.

You seem to be arguing that if you program sentience then there is no need for sentience.

 

[Vendicar]

Trees do not run from bears because the evolutionary history of plants has not included the evolution of a nervous system to be able to model the world etc.

And why would your cell phone evolve any differently than a tree?

You are conflating lots of things here. The conversations that have gone on in this thread are trying to explore whether or not consciousness is necessary for an intelligent agent or if it is just a factor of our evolution.

The title of the thread is "What is the purpose of sentience." You have confused sentience with consciousness. They are different but related concepts.

 

[D H]

Vendicar, stop with the cell phone nonsense. It is a red herring. A fallacy.

This thread is heading in the same direction as the now-locked "Limits of AI" thread, and for the same reasons: Use of terms without any solid definition and arguing based on beliefs rather than upon scientific knowledge.

I'll pose a few simple questions:
Are insects such as moths and crickets sentient?
Do moths and crickets have qualia?
Is the acoustic startle response a sign of sentience, or is it just a mechanism used by an automata to escape being eaten?

 

[GladScientist]

"Purpose" is not appropriate unless you're buying into some superior being who/that rationally established the capability.
Reference to "function" is much better and you answer your own question in description.

I did not mean purpose in the sense of "what is the purpose of a car?", rather I meant it as in "what is the purpose of a hand"? As in what biological function does it serve in our survival?

I don't have time to read every post right now but what I've read so far looks like some interesting responses, thanks.

 

[Vendicar]

"what is the purpose of a hand"?

It is difficult to keep your legs out of the bear's mouth when you aren't sentient enough to know that you have legs.

 

[Vendicar]

Are insects such as moths and crickets sentient?

Hold on... Let me ask one...

...

Nope.


On the other hand, my computer here is very self aware. It know exactly where it puts every piece of information.

However, it isn't conscious, or aware that it is aware. Neither is it aware of it's surroundings. Hence it isn't sentient.

 

[Vendicar]

Is the acoustic startle response a sign of sentience

Since insects have it and have no capacity for introspection and no consciousness the answer would have to be that it is not a sign of sentience.

 

[-Job-]

The concept of the Turing Test is relevant to this discussion.

Supposing that a computer can pass the Turing Test (and there's no evidence that a computer can't) then the behavior of a human can be successfully modeled and reproduced without sentience, to an arbitrary degree, through which we can argue that there is an evolutionary path that leads to non-sentient human-like beings.

Possibly the Anthropic Principle also applies and yet i don't see why non-sentient, intelligent beings would not at some point also debate the concept of sentience, as we are.

 

[Vendicar]

The concept of the [/PLAIN]
Supposing that a computer can pass the Turing Test (and there's no evidence that a computer can't) then the behavior of a human can be successfully modeled and reproduced without sentience,...

What makes you think that sentience can't be programmed?

On the other hand what makes you think that it can be emulated to the point that Turing can be passed without actually producing sentience?

The case for programming sentience is far stronger.

 

[GladScientist]

Hold on... Let me ask one...

...

Nope.


On the other hand, my computer here is very self aware. It know exactly where it puts every piece of information.

However, it isn't conscious, or aware that it is aware. Neither is it aware of it's surroundings. Hence it isn't sentient.

I wouldn't call those things requirements for being sentient. Obviously, the reason that my computer isn't aware of its surroundings is because it doesn't have any sensory input from the outside world. What about the Wii? It's made to be able to sense things outside of its own vessel. If it could sense everything outside of itself, would that make it sentient? I say no. As Ryan puts it, the ability to feel, make decisions, think, etc etc etc are all things that a sufficiently advanced machine could do, without being at all aware of itself.

 

[-Job-]

On the other hand, my computer here is very self aware. It know exactly where it puts every piece of information.

However, it isn't conscious, or aware that it is aware. Neither is it aware of it's surroundings. Hence it isn't sentient.

Google Goggles, an app for mobile devices, can use the phone's camera, perform image recognition, identify a Sudoku instance in the image (e.g. in your newspaper on your table) and solve it for you - all of this, in principle, without being sentient.

Have you seen the movie, The Terminator 2? (yes I'm quoting Terminator 2) In your opinion, is the terminator sentient?

 

[Vendicar]

I wouldn't call those things requirements for being sentient. Obviously, the reason that my computer isn't aware of its surroundings is because it doesn't have any sensory input from the outside world. What about the Wii? It's made to be able to sense things outside of its own vessel. If it could sense everything outside of itself, would that make it sentient? I say no. As Ryan puts it, the ability to feel, make decisions, think, etc etc etc are all things that a sufficiently advanced machine could do, without being at all aware of itself.

And none of which are equivalent to sentience, although they are closely related.

If you are defining sentience as simply consciousness then you have a problem because there is no active definition of consciousness.

But the question wasn't about consciousness, but sentience.

Specifically "what is the purpose of sentience." and that has been answered. It is an evolutionary path to the development of high level strategies needed to compete in an environment dominated by environmental noise.

 

[Vendicar]

identify a Sudoku instance in the image (e.g. on in your newspaper on your table) and solve it for you

Wake me when the telephone is aware of what it is doing.

Have you seen the movie, The Terminator 2 (yes I'm quoting Terminator 2)? In your opinion, is the terminator sentient?

Your question is really asking me of I think a machine can be sentient. It most certainly can be but without the ability to sense and respond to the world around it, you begin to run into the limits of what is commonly known as sentience.

 

[-Job-]

Your question is really asking me of I think a machine can be sentient. It most certainly can be but without the ability to sense and respond to the world around it, you begin to run into the limits of what is commonly known as sentience.

The discussion is focused on the utility of sentience. If Goggles can beat me at Sudoku and the non-sentient T2 can terminate me, then evolutionarily speaking sentience is not a trait that would be selected for because it's either not useful or unnecessary. Hence, why are we sentient to begin with?

 

[GladScientist]

And none of which are equivalent to sentience, although they are closely related.

If you are defining sentience as simply consciousness then you have a problem because there is no active definition of consciousness.

But the question wasn't about consciousness, but sentience.

Specifically "what is the purpose of sentience." and that has been answered. It is an evolutionary path to the development of high level strategies needed to compete in an environment dominated by environmental noise.

What is the difference between consciousness and sentience? I think that consciousness is just sentience, including other things (such as actually being awake and alert).

As for sentience, we make our decisions before we are actually aware of them. So how can it possibly used for strategy/decision making?

 

[Vendicar]

As for sentience, we make our decisions before we are actually aware of them.

Largely true, but as an absolute truth it remains supposition, and even if true doesn't alter the fact that feedback from the conscious can alter the next decision to be made.

I look at a clock. Note that it is 12:00 and become hungry as a result.

Conscious influence of the subconscious?

 

[D H]

Thread locked pending moderation.