Moving on from system states (and states of the system), today I’d like to fly over the landscape of different systems. In particular, systems that are — or are not — viewed as conscious.
Two views make this especially interesting. The first holds that everything is computing everything and — under computationalism — this includes conscious computations. The second (if I understand it) holds that anything that processes input data into some kind of output is conscious. (I’m not clear if the view also sees an input-output system as a computer.)
So I want to explore what I see as major landmarks in the landscape of systems that… well, about the only thing we can probably all agree on is that they do something.
I’ll start a list of putative ways to generate human (or human-like or high-level or whatever, you know what I mean) consciousness:
- The Human Brain
- A “Positronic” Brain
- A Physics Simulation (p-zombies?)
- A Neural Net Simulation
- A Functional Emulation (b-zombies?)
- (something new)
We know consciousness arises in members of group #1.
Or, per skeptics, we know something happens (something that is often casually and, in many opinions, sloppily or even incorrectly, named “consciousness” — probably by fools who don’t even know what’s happening inside their own heads).
Whatever that something is, it seems to allow its possessors to reflect on their own (illusionary) thought process in addition to processing (possibly illusionary) thoughts about the world. It’s even given them the idea they have free will (as if)!
This something allows members of this group to build imaginative creative mental models of things that don’t even exist and, in many cases, go on to build them. Or in others, to write about them.
This something has allowed its holders to take over the planet. And possibly also break it. (Good going, humanity.)
This something also seems connected with language (especially literature) and art and music and mathematics.
This something, whatever it is, is pretty fucking amazing.
So what about group #2 on the list?
This features a “brain” that replicates the structure and behavior of a human brain. It has “neurons” and they are connected with “synapses” — albeit possibly with a larger population, more connections, or faster operation.
Given a sufficiently similar architecture and behavior, the question seems more why it wouldn’t work. Why would biology matter?
If we replicate all the properties of the human brain, except its biology, it would be startling to discover the key properties were in the biology.
Let’s assume Positronic brains work approximately like ours and, therefore, the FA something arises for them, too.
That brings us to #3, a physics-level simulation of the human brain.
This is one of the few strong arguments I see for computationalism. Given a simulation of the physics of the brain, why wouldn’t it produce the same results as the brain.
An important question involves the level of the simulation. There are many choices:
- Quantum — simulate the fabric of reality.
- Atomic — simulate the atoms.
- Chemical — simulate the valence electrons.
- Molecular — simulate the compounds.
- Cellular — simulate the basic bio-machinery.
Above cellular we get into level #4, which comes next. Note that, in all of these, the brain doesn’t exist, as such. These simulations could just as easily simulate a pot roast (the first four could simulate a bowling ball).
To the extent that “everything is chemistry,” quantum- or atomic-level simulations might be overkill. (Which would be good because they’d be huge! And slow.)
But we don’t know what properties are important, yet, so it’s not a given that, when it comes to consciousness that “everything is chemistry.”
One thing about this group, and the next two, is that we’ve gone from the physical world to the world of numbers. Many believe this doesn’t matter, but it’s a huge difference in approach — literally the difference between analog and digital.
Kind of a Yin-Yang thing, in other words.
Now comes item #4, a simulation of the neural net.
We know such simulations can display learning and identification behavior similar to how ours does.
But the objection for a physics simulation applies here also: We don’t know what properties are important for consciousness.
They might involve real-time physical behaviors. My example is how some physical materials, in the right circumstances, emit coherent photons (laser light).
This behavior can be simulated with great precision. But simulations don’t ever emit photons of any kind, let alone coherent ones generated by a physical process of “Light Amplification by Stimulated Emission of Radiation.”
(For me, this argument kills computationalism. It says the outputs are appearances and byproducts; it’s the process that matters.)
Lastly (for now), comes #5, a functional representation of consciousness with no necessary similarity to how our brains work and no attempt to emulate or simulate its states.
On a crude level, this is old-fashioned AI, the old approach. It’s seen some success with expert and knowledge systems. And, I believe, in robotics (not an area I follow).
The arguments for, or against, computationalism apply here, plus a functional representation more strongly raises the issue of something that acts conscious being conscious.
Given the appearance comes from a different mechanism, perhaps one with no apparent phenomenal machinery, what might that suggest about our consciousness (is a question some ask).
The behavioral type of philosophical zombies live here. The standard p-zombies live up on level #3. That zombies live on these two levels might suggest something about their coherency (or not).
(Item #6 is an unknown unknown I’ll leave alone for now.)
§ § §
Now here’s another list.
- A Human.
- Lt Cmdr Data (Star Trek).
- HAL (2001: A Space Odyssey).
- My Dell laptop.
- A Relay-based PBX.
- An old-fashioned Thermostat.
We can generally agree (minus a few holdouts) that members of group #1 are conscious. (Items #1 and #2 here more-or-less match items #1 and #2 on the first list. The lists diverge from there, though.)
Commander Data, we are told, is conscious. He certainly acts conscious. And he reports having phenomenal feelings. Data appears to have a Positronic brain.
HAL seems to form an interesting border case. He goes insane, and his breakdown as Dave shuts him down seems phenomenal (What are you doing Dave?). I’d have to watch the movie again, but I wonder if it’s possible to argue HAL did not have phenomenal experience.
The crucial distinction with these robots is they do not (as far as I know) have a Positronic brain, but work by computational principles.
Whether they can have phenomenal experience is the topic of heated debate.
With some exceptions, most believe laptops do not have phenomenal experience. Specifically, there is nothing it is like to be a laptop.
I included the PBX as an example of a system with complex behaviors arising from lots of basic components (lots and lots of relays).
Note that a relay-based PBX is electro-mechanical (rather than electronic). It consists entirely of electromagnets and mechanical switches.
Lastly, a thermostat is just a temperature-operated switch. There are also light-operated, time-operated, and motion-operated, switches.
In all cases, from laptops down to thermostats, some believe there is phenomenal experience, even consciousness.
I don’t fathom that view at all. I don’t believe in Pixies or in the fundamental consciousness of rocks or thermostats or even my laptop.
I’ll explain why in the next post.
Stay conscious, my friends!
 Attitude? What attitude?? (Look, take all of this a little tongue-in-cheek. No one knows the right answers here, so I’m just amusing myself.)
 Sorry! There’s really just no other way that says it as right.
 Or they’re really hard to make and have fewer neurons, fewer connections, and run slower. They make dim-witted robots that are still capable of complex tasks.
 Much of the argument’s strength comes from the success of such simulations in other aspects of reality, in particular biology.
Its weak point is that we can’t know if the properties of consciousness arise from a simulation. The laser simulation analogy (or weather simulation analogy) attacks that weak point. It all depends on whether consciousness is in the process or the outputs.
 If not, there are other computers from SF we could use. Colossus maybe? How about Robbie? Or his pal, Robot, from Lost in Space?
 I might argue that Data, HAL, and all the rest, being science fiction speak to a real gap between #1 and #4, but, yes, I know exactly the counter-argument for that.
OTOH: The Lunar Hilton, my flying car, transporters, and FTL. 😀