Can Being Aware of the Illusion of Self Augment an Agent’s Affordances:

What to take away

1. 1. Buddhist 'no-self' doctrine, which frames the person as an imputation over impermanent causal factors rather than a substantial enduring individual (Kapstein, 2016; Garfield, 2014), is proposed as a conceptually actionable input for designing artificial agents with altered self-models.
2. 2. Mindfulness meditation practices that reduce 'selfing' are empirically linked to increased social connectedness in a 2008 Hutcherson, Seppala & Gross compassion-meditation study published in Emotion (vol. 8, no. 5, pp. 720–724).
3. 3. A Luberto et al. (2018) systematic review and meta-analysis across multiple meditation studies found consistent increases in empathy, compassion, and prosocial behaviors, constituting the paper's primary quantitative human-substrate evidence for affordance expansion.
4. 4. Lucid dreaming — defined as dreaming in which the dreamer is aware they are dreaming (LaBerge & Rheingold, 1991) — is identified as a second cognitive substrate in which self-as-illusion awareness opens motor and environmental affordances not available in waking life (Tholey, 1989).
5. 5. The formal instrument the paper introduces is an information-theoretic modeling framework combining Predictive Information (Bialek, 2001; Ay et al., 2008), Synergistic Information (Edlund et al., 2011), Multi-Information (McGill, 1954; Schneidman, 2003), integration (Tononi et al., 1994), and empowerment (Klyubin et al., 2008) to operationalize self-boundary and affordance change in artificial agents.
6. 6. Integrated Information Theory 3.0 (Oizumi, Albantakis & Tononi, 2014, PLOS Computational Biology 10(5): e1003588) and the information theory of individuality (Krakauer et al., 2020, Theory in Biosciences 139(2): 209–223) are positioned as foundational substrates for formally modeling selfhood in artificial systems.
7. 7. The artificial life approach is explicitly distinguished from AI in that it does not attempt to replicate self-as-illusion directly but models its functional effects within synthetic organisms, following the bottom-up methodology of Bedau et al. (2000) and Beer's (2014) glider cognition work in Artificial Life 20(2): 183–206.
8. 8. A replicable methodology proposed is to connect sensorimotor-loop agent models (Varela, 1997; Beer, 2014) to empowerment measures (Klyubin et al., 2008, PloS ONE 3(12): e4018) and then vary the agent's encoded self-boundary to test whether reduced individuality increases the empowerment score as a proxy for expanded affordances.
9. 9. An open hypothesis the paper raises is whether insights about self-as-illusion derived from human contemplative practice can be formally transferred to non-human minds to produce qualitatively new categories of affordances that have no analog in human cognition.
10. 10. Mahāyāna Buddhist philosophy (Maitreya-Asaṅga et al., 2014) describing self as simultaneously extendable and illusory is proposed to map onto the computational concept of distributable individuality, suggesting that agents with distributed self-models may incorporate others into their action-selection loops and thereby expand social affordances.

Peer brief — for seminar discussion

This ISAL 2021 proceedings paper, funded by a Templeton World Charity Foundation grant and authored across institutions including McGill University's Department of Psychiatry, Cross Labs Japan, and the Earth-Life Science Institute at Tokyo Institute of Technology, poses whether encoding awareness of the illusory nature of selfhood into an agent — artificial or biological — qualitatively expands that agent's affordance space. The approach is explicitly programmatic rather than empirical: no new experiments are reported. Instead, the paper constructs a conceptual bridge from three existing evidence streams — Buddhist no-self philosophy (Kapstein, 2016; Garfield, 2014), contemplative neuroscience (Britton et al., 2013; Hutcherson, Seppala & Gross, 2008 Emotion study; Luberto et al., 2018 meta-analysis), and artificial life modeling (Bedau et al., 2000; Beer, 2014 Artificial Life 20(2)) — and proposes a formal instrument to test the bridge: an information-theoretic framework combining Predictive Information (Bialek, 2001), Synergistic Information (Edlund et al., 2011), Multi-Information (McGill, 1954), and empowerment (Klyubin et al., 2008, PloS ONE 3(12): e4018) as operationalizations of self-boundary and action possibility. Lucid dreaming, documented phenomenologically by Stumbrys et al. (2014) in American Journal of Psychology 127(2), serves alongside mindfulness as the human-substrate proof of concept that reduced self-reification unlocks previously unavailable affordances. An alternative method that could have been used is direct agent-based simulation: evolving animats under different self-boundary parameterizations and measuring empowerment scores, an approach Beer (2014) and Edlund et al. (2011) have used in adjacent problems. The load-bearing finding is that 'selfing' — the reification of a stable, bounded individual — functionally constrains affordance repertoires, and that any system (human meditator, lucid dreamer, or artificial agent) that implements a representation of self as distributed and illusory will exhibit measurably expanded action possibilities, potentially including prosocial and inter-agent affordances with no equivalent in conventional agent architectures. The paper predicts this will generate novel foundational principles in AI science and new technology categories. The hypothesis is that awareness-of-illusion, not merely reduced self-boundary, is the operative variable — the metacognitive layer matters, not just the structural one. A critical reader would push back on the most fundamental methodological gap: the entire argument rests on analogical transfer from contemplative phenomenology and meditation meta-analyses to artificial agent design, with no demonstration that 'awareness' in the required metacognitive sense is even computably defined for current artificial life substrates. The empowerment measure from Klyubin et al. captures options available to an agent but says nothing about whether the agent represents those options as arising from a self-model — the conflation of structural self-boundary relaxation with genuine awareness-of-illusion is the paper's central, unresolved inferential step.

Claims (3)

• Testing theories of self-illusion requires dual process: using knowledge of human self-experience while testing effects in human and other minds.

  Author's methodological assertion: integration of contemplative insight with computational modeling.

• Affordances are relationships between the abilities of organisms and features of the environment.
• The self can be considered both extendable and illusory in Buddhist thought.

  Author assertion: rejection of singular self enables perception of distributed/dynamic selfhood; foundation for affordance augmentation hypothesis.

Similar preprints — Semantic Scholar