Westworld’s Bicameral Mind hypothesis

Westworld, no doubt, is impressive in terms of production quality and intricate storytelling, which results to a highly addictive, intelligent and thrilling drama. It also makes clever references to current research in the technological, psychological and philosophical fields. Although, this might just serve as an intellectual smokescreen, it indeed raises profound questions about machine intelligence, artificial consciousness, and the means to reach them.

The show has grounded its approach to synthetic consciousness in the theory of Bicameral Mind. Initially, this was a theory of human consciousness development. To clarify, biological consciousness could be broadly divided into three areas: phenomenal and reflective consciousness, and self-awareness. Phenomenal consciousness is one’s raw experience of colours, shapes, sounds and feelings. Reflective consciousness is basically introspection, which involves access to one’s innate knowledge, memories, occurrent thoughts, allowing to manipulate own behaviour. Lastly, self-awareness is the recognition of the self as a different agent to the environment, recognition of own phenomenal and access awareness and the ability to conceptualise these abilities. Arguably, self-consciousness characterises subjectivity as such, serving as a link between experiences and first-personal accessibility to them. Therefore, introspective consciousness would not be possible without recognition of own subjectivity.

The Bicameral Mind theory attempts to explain how humans have reached introspective self-awareness. In 1976 book The Origin of Consciousness in the Breakdown of the Bicameral Mind, Julian Jaynes made the case that modern human consciousness only emerged approximately 3000 years ago. Jaynes studied ancient writings and noticed that the characters of, for example, Iliad or the Old Testament lack introspection or any subjectivism. Instead they followed the commands of the external voices, known at the time as ‘gods’. In contrast, Odyssey characters reflect on their motivations and actions. His theory holds that humans used to have bicameral or “divided” minds, where one hemisphere was ‘speaking’ and the other would obey. Thus, the now mostly dormant language areas in the right hemisphere would activate in forms of auditory hallucinations, which would transmit the left hemisphere its experiences and memories. This phenomenon could also explain schizophrenia, as many schizophrenics do not simply hear random voices, but experience ‘command hallucinations’, just like the ancient bicameral-minded individuals. Humankind therefore has transcended the bicameral mind framework due to its limited cognitive capacity to cope with the complex and stressful situations onset by human development, travel, and increased population size. He concluded that the emergence of modern reflective consciousness was a cultural rather than a biological adaptation, comparable to a change of software rather than hardware.

This theory offers a compelling narrative and insight; however, it was largely ignored in the scientific field due to a number of inconsistencies. Nevertheless, Westworld writers have taken this idea to portray unconscious robots, acting upon their routines and following the commands of their ‘gods’ or programmers. Then they started getting the glimpses of their old memories in a form of visual-auditory hallucination along with audio commands from Arnold, which marked the beginning of their journey to introspection and self-awareness. The host, therefore, must have had a base ‘Host’ code, which worked as a left hemisphere, and the ‘Arnold’ code which played the role of the right hemisphere.Consequently, the hosts would follow their base commands or ‘loop’ until they are encountered with new circumstances or guests, which could inject new commands. As per philosopher Daniel Dennett, this would create new “logical spaces”, where exposure to increasingly complex language and situations bootstraps self-awareness. Consequently, when Dolores finally recognised that it was not the voice of Arnold addressing her but her own thoughts, she finally acknowledged her subjectivity, allowing to break out from the loop.

Now the question stands: Could the bicameral mind theory work as a more elegant approach for creating artificial self-aware mind? To make this question more clear, this shall be referring a different kinds of AI as it is possible to be highly intelligent without being conscious, by using divergent thinking styles or algorithms. For example, a sufficiently intelligent AI could pass the Turing test without actually having any self-awareness. This is among the main arguments of the Chinese Room thought experiments. Conceived by John Searle in 1980, it holds that an AI could imitate complex human behaviour by learning language pattern-matching algorithms to participate in conversations without realising what it was actually saying. Similarly, the early hosts of Westworld were able to pass Turing test without having any self-awareness.  This therefore would make the bicameral hosts a kind of philosophical zombies - completely devoid of self-awareness yet able to persuasively and intelligently.

As such, basing a conscious AI on bicameralism contradicts the current mainstream approaches to AI building. The artificial intelligence is usually coded as a whole, and it’s learning and further development comes as a side effect from interaction with the outside world. Yet, in Westworld the AI would have to be learning from itself. As such, it can be assumed that the hosts have a recursive algorithm, which breaks down complex tasks into smaller, more manageable ones until a predefined base case is reached. These small problems are then solved and recombined together to address the matter at hand. Having a bicameral model at work would mean that in solving these problems, the ‘Host’ and the ‘Arnold’ codes would have to be constantly communicating. The former would be inquiring the latter on how to break down the task more efficiently, and thus learning how to action independently by altering its code in the process. This means that the synthetic consciousness cannot be built but has to grow or evolve into it, like humans did.

It is yet hard to hypothesise whether it is possible. Marcel Kuijsten from MIT suggested that a robot, with sufficiently complex language, may actually become conscious if it has ”the metaphorical basis of the external world to build up an interior space”. However, it must be acknowledged that building such code must be a daunting and enormous effort. The humankind had thousands of years of evolutionary trial and learning embedded into our subconscious algorithms(instincts and reflexes), guiding our behaviour. Mimicking that in order to construct an automated inner ‘socialisation’ engine may lead to very reductionist and incomplete results. Moreover, it is often argued that consciousness may have simply been an environmental by-product or a nature’s accident. Therefore, this further decreases the likelihood of this theory working, especially considering the difficulty of recreating the conditions the human kind has gone through to bootstrap self-awareness. Finally, the hosts in Westworld also had to learn in a very specific way to be finally able to disobey or ‘transcend’ its core programming. Not entirely impossible, but yet incredibly difficult and extra encouragement on the human part might be needed.

To conclude, the Westworld idea on stimulating artificial consciousness is definitely not without substantial theoretical backing. As such, this piece attempted to contemplate the real-life possibility of such scenario. Sadly, at this point it is yet hardly possible to reach any conclusions. The modern science is not even close to capturing or even understanding the nature of conscious experience. Although, it is certainly very interesting to explore the possibilities! The thought of Bicameral recursive approach is certainly entertaining and potentially with utility. At present, there is little to none research in this direction, but this may change soon. And it is in our aspiration to be at the forefront of technological progress, understand and explore all the potential and novel ways move forwards. In the end of the day,  moon landing, nuclear power and teleportation have also once been just pieces of science fiction.