Defining life, defining sentience

We believe that having solid definitions of life and sentience (consciousness) is essential for the success of research into these areas. However, only life on Earth as we know it serves as a source of examples, including its edge cases (such as viruses). Alternatives originating in fiction can be used as thought experiments.

As long as there are only limited kinds of examples available for generalisation to be based on (namely, life on Earth), a very detailed and comprehensive definition seems to be difficult to establish. Even NASA's definition of life sticks decidedly to that which we know, “a self-sustaining chemical system capable of Darwinian evolution”.

Therefore, we at FINAL Labs™ are only putting forward a general idea about how we approach the definition of life and that of sentience.

The FINAL Labs™ Life and Consciousness Definition Framework

In our terminology, “definition” means a set of criteria that helps decide if an observed phenomenon is a manifestation of life and of consciousness.

In our framework, the response to such an inquiry is an interval of probabilities for a ‘yes’.

Just as an example, while the author of this text might consider themselves alive and conscious, convincing the reader about these will not go beyond the reliability of a Turing test.

We can nevertheless enumerate a number of factors that can contribute to a ‘yes’ or a ‘no’ answer in the assessment of the specific phenomenon. A duck test; the ability to reduce entropy; being able to reproduce; spatial propagation; adaption to the environment; sensing and acting on the environment; having an agenda; agency; homeostasis; and so forth.

The significance of each such criteria can be expressed as a weight (or rather, a weight range). As an example, the ability to reproduce would probably be assigned a smaller weight compared to many other factors. Both donkeys and mules are considered to be alive, but the mule cannot reproduce.

The various natural and even fictional edge cases can help assign a range of weights to each of these factors and thus yield a range of probability for a yes or a no.