Frequently Asked Questions
What is computational functionalism (CF)?
A formal definition is often high-level in nature, e.g. "performing computations of the right kind is necessary and sufficient for [phenomenal] consciousness" (e.g. Butlin et al., 2023), where 'performing' is taken to mean implementing in any physical substrate capable of applying the right mechanisms and 'computation' has a range of contested definitions.
For this website we operationalise a more specific version of CF that is: "a complex algorithm* gives rise to a conscious entity capable of complex conscious experiences when that algorithm is implemented physically on a digital computer, where the nature of that entity and its experience is wholly defined by the algorithm – and all possible types of conscious entity/experience are also only conscious because they share functionality that maps to the relevant algorithm".
* Or family of algorithms; which can be adequately represented on any Turing Machine equivalent architecture. The algorithms themselves are not currently well-defined under any mainstream CF positions. The 'complex' condition is introduced to navigate this uncertainty. While 'complex' is not formally defined, it is intended to exclude the possibility of trivial moments of consciousness or fleeting selves under panpsychist or near-panpsychist interpretations of CF. For instance, it excludes the possibility that every boolean AND logic operation generates a conscious entity that existed for the duration of that operation, experiencing the contents of the two inputs to the operation. Loosely, we can consider 'complex' to require at least many hundreds of basic CPU operations to implement the target algorithm and more likely millions or trillions.
Are any arguments deliberately excluded?
This website aspires to be comprehensive barring known exception classes, but remains work in progress. We note that many arguments have been made in different ways, but collapse into the lists provided. Exception classes:
- Historical arguments that are arguably adequately resolved by the capabilities of modern AI systems or by modern definitions of computation, such as the triviality critique (whether most physical systems can be interpreted as implementing any imaginable complex CF algorithm), Hume's Missing Shade of Blue (how to infer/extrapolate something that has never been seen in input data), the Frame Problem (how to identify which circumstances matter for addressing a given problem out of infinite possible ways of framing those circumstances), or capabilities previously argued as uniquely human that are contestably necessary for consciousness (e.g. language, creativity, reasoning etc.). These exclusions remain under review.
- Biological conditions which might be necessary for consciousness, such as certain aspects of metabolism, thermodynamic processes, autopoiesis, low entropy homeostasis, etc. These are excluded since we are not yet aware of strong positive arguments for any motivated function associated with those biological criteria that cannot be replicated in principle in an algorithmic system. However, these could be added in the future.
What is consciousness?
In casual conversation, it's common for 'consciousness' to refer to different things in different contexts. But this polysemy is nothing unusual (consider the words 'community' or 'space'), it just means we need to be careful to define it for specific contexts. In particular, having multiple definitions doesn't mean a concept is hopelessly confused or that reasoned discussion about it is impossible.
The formal philosophical definitions for 'phenomenal consciousness' (our scope for this website) can feel circular, although they often get close enough for good faith discussions. A system is conscious if there is an experiencing subject, some locus of awareness, or something that 'it-is-like' to be that system from the inside. Such experiences might be very simple and fleeting, without the sophisticated sense of temporally-continuous self and world model that accompanies ordinary consciousness in waking human adults. A practical definition of consciousness is what goes away in deep sleep or anaesthesia, returning again when we awake. This has some limitations as a definition, but helps point us towards the right phenomenon.
The definition we favour is a lowest common denominator approach, based on affirmative ostension. Put simply, phenomenal consciousness is the feature(s) shared across all states which you've been aware of, whether those states involve cognition, sensations, dreams, transcendent experiences, or anything else. We can never know anything or experience anything except through a state of consciousness. The more states you can experience and rigorously analyse, the closer you'll get to the reality behind this definition, but most adult humans should be able to operationalise this approach adequately to support a practical dialogue.
With language, as with any other tools we use to analyse the world or communicate, the requirement is a good enough working approach to make progress, rather than a perfect solution beyond any nit-picking.