Dec 05 2016

Westworld and Consciousness

westworld1The season finale of Westworld aired last night, a series based on a Michael Crichton book which was made into a 1973 film. I won’t give much away, so only very mild spoilers for those who haven’t seen it. I will say the last episode was probably the best of the season.

The basic premise of the book/film/series is that it takes place in a futuristic theme park in which guests can visit the old west populated by robots that are there solely for their pleasure.  They exist to lose gunfights, for sexual pleasure, to be victims or fill whatever role the guests want, and then be recycled to run through their plot loop all over again.

The HBO series uses the story line as an opportunity to explore the basic question of sentience. The robots are hyperrealistic. Unless you cut them open, you cannot tell them from a living human. They are extremely realistic in their behavior as well.

The robots clearly have a very advanced form of artificial intelligence, but are they self-aware? That is a central theme of the series. They have complex behavioral algorithms, they can reason, they express the full range of human emotions, and they have memory. They are kept under control largely by wiping their memory each time they are repaired, so that they don’t remember the horrible things that were done to them.

Some of the robots, however, start to break out of their confines. They “feel” as if they are trapped in a recurring nightmare, and have flashes of memory from their previous loops.

The show explores some interesting questions raised by this scenario. The first is, how can we tell if the robots are truly self-aware. Are they experiencing their own existence or are they the equivalent of Chalmers’ p-zombies, with no subjective experience. Chalmers’ position is that a machine can be made to do everything humans do without subjective experience, and therefore our current models of reality don’t explain why there is subjective experience.

I think Chalmers is wrong, and prefer Daniel Dennett’s approach. Consciousness is an endless internal circuit, the brain talking to itself and incorporating information derived through perception. Subjective experience, Dennett argues, is inseparable from the processing that our brains do. Behavior motivators, the basic reward and aversion circuits that are fundamental to central nervous systems, have to feel like something. Also, existing needs to feel like something or otherwise we would not be able to tell the difference between a memory and a live experience.

I found that one interesting tidbit in Westworld – the robots can sometimes immerse themselves in a memory and cannot tell the difference between a memory and a live experience. They can relive prior memories. This is actually a detriment, because it is confusing and they sometimes cannot tell if they are experiencing the current moment.

The designers of the robots in Westworld are struggling with the question of if their creations are sentient. Mild spoiler here – one programmer realizes at one point that the last critical ingredient to making the leap from software to sentience was having an inner voice, being able to talk to oneself. Other elements, like memory, perception, and behavior, were necessary but insufficient. They had to close the internal loop. That is probably not a bad metaphor for human sentience.

I also like that the show confronts another issue, the notion that the robots are completely programmable. How does that affect whether or not they are sentient. One of the robots that seems to have bridged the gap to full self-awareness is confronted with the fact that her behavior can be programmed. She resists the implications of this, that she has no real free-will, and insists that she is making her own decisions.

The show is just starting to confront that issue, but I think it has already delved pretty deeply into the question of sentience and free will. That question is beyond the scope of this article, but to quickly summarize, some philosophers feel that the deterministic aspect of how our brains work mean that they must follow the laws of physics, which means we don’t really have any free will. Our thoughts and behaviors and the inevitable outcomes of physical processes playing themselves out. This does not preclude making decisions, but those decisions follow the laws of physics.

The robots of Westworld are no different than humans. It is just a lot more obvious that they can be programmed, and you can do so directly if you have access to their software. In essence, they are machines. But humans are machines also, and we can be programmed. The only difference is that we don’t currently have the technology to fully access and control our “code.” There is no scientific reason, however, why we will not someday have access to that code. It is also very likely that we will merge our brain function with machine intelligence, but that is yet another topic of discussion.

Psychologists, in a way, study how to hack into the software shared by most people. It is clear that human behavior can be statistically influenced by external factors. In essence you can strongly influence the perceptions, beliefs, and behaviors of others if you know how how to hack the brain’s algorithms. This is essentially what magicians do, and there is increasing overlap between psychological experiments and magician performance. Mentalist Darren Brown has essentially built a career out of converting psychological experiments into magic tricks.

And yet people generally resist the notion that we are so easily manipulated. It does not feel like we are being manipulated. We feel like we are making a free choice, that we have reasons for what we do. But those feelings are just part of the brain’s algorithms, they are just another loop of the brain talking to itself and creating the illusion of free will. We are generally not aware of most of the things that influence our behavior. When asked to explain our behavior, we do not recognize even elements that were deliberately placed to influence our behavior, and instead will make up internal reasons.

The most dramatic examples of this are in the split brain experiments. If you cut the corpus collosum between the two hemispheres of a human brain, you end up with two separate brains each with their own consciousness, but sharing a body and still with some limited communication. If, say, you show one hemisphere an image and it then acts on that image, and then you ask the other hemisphere to explain the behavior, it does not know. It cannot know because it was not privy to the image. The person, however, still sees themselves as a single entity and is not aware of the duality in their split brain.

So, the left hemisphere must explain what the right hemisphere just did, when it cannot know. The person, however, does not say that they don’t know. They provide a plausible explanation for the behavior, one that we know is wrong, but the person believes it was their motivation. They were just as “programmed” as the robots on Westworld, and just as sure that they are making their own decisions.

The bottom line of all this, in my opinion, is that the fact that the robots of Westworld can be completely programmed is not incompatible with the conclusion that they are fully sentient and self aware. They are sentient machines, and those machines can be hacked. Humans are sentient machines, and our machines can be hacked as well. Right now we mostly hack our brains through perception inputs. We also hack them chemically. When necessary we can hack them surgically. We are just starting to hack them electrically.

Eventually we will be able to hack them at a neuronal level, Matrix style, and we will truly be no different than the robots of Westworld.

 

60 responses so far