Rog’s Day 2: Enter the Matrix
How Simulations might be Sufficient to Generate Consciousness
It actually started raining. Tic. I heard seldom drops on the window overhead. Tic. I tried to ignore them — tic tic — with no avail. I started involuntarily focusing on the drops. Tic. Waiting for the next. Tic tic. Trying to anticipate when the next would fall. Tic Tic. Tic. Realising that the intervals were getting shorter… Tic tic… Tic tic tic tictic tictictictictictic tictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictictic
The heavy rain made it impossible to distinguish the individual drops, and the white noise catapulted me back into the room. I suddenly felt cold, and I only realized how much time had passed by hearing which song of the playlist was on. Many sets of the same 4 chords have been played, without me hearing any of them. This must have been going on for a while, but the drops had taken all of my attention. And I was starving!
While hearing was the main sense I was conscious about, there was a lot more going on. I was having a sense of the time between drops, and trying to extrapolate a pattern in order to anticipate. My sensation of cold was amplified despite the temperature not changing significantly. And even before it started, I became conscious that it was about to rain heavily.
I was using just one external sense, and that was sufficient to become conscious about several aspects. The question that arises naturally is:
Is one sense sufficient to develop consciousness?
There are no conscious beings having just one sense (or even 2), therefore there’s no simple answer. On the other hand, in an effort to create a conscious robot, reducing the complexity of the system is so valuable that it makes it worthwhile digging deeper into the question.
What is a sense, anyway?
Biological senses are organs that transform a physical quantity into data. Data is coded in many ways, and is then processed into percepts.
Albeit similar and too often used interchangeably, sensing and perceiving are very different. Illusions are a powerful way for understanding their difference. For instance, in peripheral drift illusions, we use sight to look at a static image, but we perceive motion.
Sensations are representations of reality,
while percepts are interpretations.
The interpreted representations depend on the influence of other senses, as well as the structure of the brain. For instance, smelling fresh bread makes you feel more hungry. Or looking at snowy winter scenes is likely to make you feel colder. On the other hand, even within one sensory modality, different parts can modify the percept. Like in the drift illusion above, motion perception is created within vision only, because of differences in how processing occurs in the periphery of the field of view.
These observations yield 2 consequences:
(1) Unifying Coding. The fact that different senses can interact with each other to create percepts means that the brain is able to unify data into a common field (which is not necessarily homogeneous in coding, and not necessarily source-agnostic) [We call it Cognitive Playground].
(2) Intrasensory Interactions. Additionally, there is interaction even within the field of a single sense, leading to changes in perception.
If on one side this seems obvious (How’d the brain process information otherwise?), it allows us to set a reductionist hypothesis; In other words: Consciousness might arise even from one single sensory modality.
This idea is highly convenient in terms of implementation, because it facilitates the creation of a testing environment. Instead of having to control all variables that might affect all the senses, we just need to handle one sensory dimension. This reduction in dimensionality could mean that the problem becomes more tractable, and actually lead to an answer.
The caveat is that even a single-sense world can be utterly complex — just look around you and marvel at every little detail of the real World!
Complex in terms of interactions between entities in space (e.g., a cat being hovered above ground by a table) and also in time (how the cat will definitely make the glass fall from the table). Complex because these interactions behave differently at different levels of observation (advocates of a quantum-based explanation of consciousness combine several of these levels). The world required for consciousness to emerge might be so complex that the solution space is too large to explore. That is, unless we control it, which we can do by either constraining the physical world or creating a virtual reality. The latter is clearly faster and more scalable: A simulation allows to control and grow complexity little by little, testing out hypotheses on how consciousness arises at each step.
Some are skeptical of this approach, convinced that a simulation will never be the real thing. “Algorithms can predict the weather, but it’s never wet inside the computer.” Of course not, but that’s not a very strong argument against it either. Therefore, we take our chances and start building. As said at Rog's Day Zero, we might learn something out of it.
Of course, we are aware of the limitations of this approach. For instance, a consciousness that arises from a single sensory modality cannot be conscious of what other senses feel like. Like hearing for the first time or seeing for people who were born blind (recovery from early-age blindness is similar, and shows how the brain can cope with a flood of “new” sensory information). Consciousness of a sense requires an internal representation. But we start with one, which still appears like an impenetrable wall.
Meanwhile in the cave
We left Day 1 with the question whether a simulated world would be sufficient to enable the generation of a conscious being. The answer is yes, provided that the simulation is sufficiently rich, and the “being” has sensory, processing, and motor capabilities that match the content of its world.
We were also in Plato’s cave, tied since childhood, forced to watch eerie figures dancing on a screen. Now we start to understand that there might be more, that what we see is not all there is to it. Now it’s time to start to move.
