‘What is Life’ and ‘What is Mind’ are simple questions lacking simple answers; there seems an unbridgeable gap from the lifeless, mindless world of physics, perhaps prompting appeals to the supernatural or to ‘Ghosts in the Machine’. Artificial Life researchers have to acknowledge such basic philosophical issues at the core of their studies; but they should reject the notion that living, mindful beings are supernatural. Following Ryle and Wittgenstein we argue that the simple questions are too often made nonsense through category errors, confusion between different levels of description. We consider the role of the observer, and the opposition between objectivity and subjectivity. The sciences of life and the mind are particularly prone to confusions about the roles of observers. We highlight the practical consequences for Alife research. Such category errors undermine the rationale for large bodies of this research. We discuss how to identify these suspect assumptions and how Alife research can avoid these traps.

Maps are useful for navigation if (i) there is adequate known detail provided on the map, (ii) your present location on the map is known as is (iii) the location of your goal. Many natural examples of successful navigation, such as seasonal bird migration across continents and oceans, lack some or all of these. Success requires some strategy for continuously governing the direction of movement according to continuous sampling of available sensory cues. If sensory cues are strictly local, an instantaneous snapshot often gives insufficient guidance on the course to steer. Some form of active perception is needed, where motion gives rise to cues as to whether the current course needs to change. We illustrate with one old and two novel examples of increasing complexity: (A) Run-and-Tumble strategies, such as used by e. Coli , allowing the detection of local gradients. (B) Swerve-Zone, a novel artificial life model of bird migration showing how region-wide cues, even in the absence of discernible local gradients, can nevertheless still guide. And (C) Parting-the-Waves, a proposed strategy for exploiting the wave patterns underlying long-distance steering as used by Micronesian navigators, showing how the boat motion is essential for discriminating between swells. All three depend on some default motion; when stationary, you cannot sense where the goal lies. They exploit motion in different ways, but all simplify navigational search into tractable forms apparently amenable to evolution.

In my talk I give a broad-brush overview of 4 billion years of Real Life, and 2000 years of Artificial Life. The role of AI, as a subtopic of ALife, is discussed in the context of the early (1950s on) AI distinction between Neats and Scruffies. Roughly, the Neats believed that our reasoning abilities depend on reasoning “all-the-way-down”, they saw computers as the model for the brain, and this led to the dead-end of GOFAI. Scruffies believed humans and animals were more ad hoc machines crafted by evolution; the ability to reason is an explanandum (to be explained), not an explanans (taken for granted). I give my updated version of this distinction: when trying to explain (or re-create) some high-level regularities in biological phenomena, the Neat approach is to assume these echo somewhat similar low-level mechanisms; whereas the Scruffy approach sees no need for such rule-bound assumptions. Holograms are introduced as a powerful metaphor for the Scruffy approach, showing how high level labels for cognitive phenomena can have zero direct low-level correlates in the underlying machinery. Attractors and invariants in physical and social dynamics are key to understanding life and cognition. I give a historical connection between such holographic principles and Deep Learning, a resounding triumph for Scruffy AI -- though still not Scruffy enough. Further connections are drawn between holograms and recently proposed explanations for Micronesian wave navigation. ALife is much broader than AI, and needs to go deeper still; depth in space and time is discussed. There are boundless opportunities for innovative ALife research, provided one follows the Scruffy approach of challenging unsupported assumptions. The so-called Singularity is characterised as a fantasy of the Neat approach taken to its absurd limits, that misdirects attention from the very real political and social dangers of AI. In contrast, the awe-inspiring Monarch butterfly illustrates the spectacular range of biological phenomena that ALifers should be celebrating.

Constructing an autonomous robot or artificial agent using an Artifical Life perspective requires analysis of its perceptual interface with its world. Appropriate sensorimotor dynamics are needed for its embodied interactions with the physical environment in which it is embedded. Similar issues occur for small craft navigation across the seas. These parallels are explored in the context of a newly proposed explanation for the dilep , a wave-mediated pathway between islands in the Pacific as used by traditional wave-navigators in the Marshall Islands.

Life and cognition are inherently circular dynamical processes, and people have difficulty understanding circular causation. We give case studies illustrating some resulting confusions, and propose that the problems may lie in failing to properly distinguish between similar concepts used in different levels of description of a system, typically local and global levels.

In both social systems and ecosystems there is a need to resolve potential conflicts between the interests of individuals and the collective interest of the community. The collective interests need to survive the turbulent dynamics of social and ecological interactions. To see how different systems with different sets of interactions have varying degrees of robustness, we need to look at their different contingent histories. We analyse abstract Artificial Life models of such systems, and note that some prominent examples rely on explicitly a-historical frameworks; we point out where analyses that ignore a contingent historical context can be fatally flawed. Real life studies highlight the role of history, and Artificial Life studies should do likewise.