Source
Michael Levin (2019), “The Computational Boundary of a ‘Self’: Developmental Bioelectricity Drives Multicellularity and Scale-Free Cognition”, LIT-439, Frontiers in Psychology 10: 2688, read in full in NOTE-341. The cited experiments were not checked against their sources.
What was actually shown
The definition. “The scale and types of goals that a system can pursue defines (determines) the boundaries and content of the putative ‘agent’.” The boundary is the “cognitive light cone”: “the most distant (in time and space) set of events that this system can measure and attempts to regulate”. A goal is the set point of “a feedback system that operates to maximize some specific state of affairs”. Fig. 2 places a tick, a dog and a human on a time × space diagram, called “semi-quantitative”.
Nesting. “Individuals can overlap – the same biophysical system can support a number of coexisting, coupled Selves with different cognitive borders.” Swarms, animals, organs, cells and intracellular networks “each have their own cognitive horizon”. “There is not necessarily one unique primary level of organization”; the level to describe is the one giving prediction and control “with the least effort”.
Binding by shared information. Cells coupled by gap junctions share measurements, so the coupled group can sense and regulate what no cell can see, such as the length of a limb. The cells stay “equally selfish”, with “the relevant self that they defend … bigger in terms of space, time, and complexity”. An inner agent keeps its own Self “only to the extent that it regulates (restricts) information flow from neighbors”. Too little sharing fails to bind, and too much gives “a homogenous soup”.
Cited evidence. Left–right identity is randomized by whole domains, not cell by cell. Depolarizing a specific cell population makes normal melanocytes metastatic in genetically normal tadpoles, and hyperpolarization prevents tumours from human oncogenes. Plastic barriers and gap-junction blockers induce tumours. Most anaesthetics are gap-junctional uncouplers, offered as “no coincidence”.
What could have come out otherwise. Changing bioelectric coupling could have left pattern and tumour outcomes unchanged. In the cited experiments it did not.
What this does not say
- That the boundary can be measured. The light cone has no units or procedure, and “much more work is needed to fully flesh out this rubric”. It cannot yet decide a colony, a holobiont or a nation.
- That a goal-pursuing whole is anything more than a homeostat. Goals are ascribed: “It is of course non-trivial empirical work to ascertain goals.” On this definition any homeostat is a Self, which the paper accepts.
- That the experiments show selves. They show that voltage and coupling change pattern and tumour outcomes. That they change the boundary of a Self is the paper’s interpretation.
- That IIT is wrong. The contrast is one bracket aimed at Tononi 2008. Levin’s nesting is about goal pursuit; IIT’s exclusion is about experience. The two are not yet in contradiction.
- That tissues or groups experience anything. “Nothing in this model explains” the Hard Problem. The suggestion that there is “something-it-is-like to being a tissue or organ” is a closing speculation.
- How the larger agents evolved. The infotaxis and surprise-minimization scenario is not modelled or tested.
Connections
- QUESTION-009. It gives a partial answer: a pattern of coordination is an additional agent when the shared information lets the parts pursue a goal none of them can sense. It is liberal, because any homeostat qualifies, and it has no measurement yet.
- Minimal agency (THEORY-071). That account denies that individuality alone, a homeostat’s included, is agency, and requires interactional asymmetry and self-generated normativity. Levin individuates by the goal, THEORY-071 needs the individual before the goal. The two disagree about what makes a goal-pursuing collective an agent, and they cannot both be right about the homeostat: here it is a Self, there its externally fixed norm makes it no agent. So they are rivals.
- Minded parts and minded wholes (THEORY-043) already uses this paper as the composition-side statement for agency, and LIT-159 (NOTE-131) is its counterpart for experience.
- Collaboration and closure (THEORY-418). That sketch also puts a shared goal at the centre of a higher-level whole, but requires collaboration, entrenchment and organizational closure, and makes the whole an individual rather than an agent. Levin’s criterion asks less.
- A comparative criterion. THEORY-232 (Watson, LIT-990) says when a whole is more agential than its parts, without a threshold. Levin’s criterion is a threshold of the kind THEORY-232 declines.
- Causal emergence (LIT-025). Levin names Hoel’s measures as the way to find the most efficacious level; using them would turn his pragmatic level-choice into a measurement. Whether they pick the same levels as the goal criterion is open.
- Dennett (LIT-220). The level-choice rule is the intentional stance.