I think I got onto this train of thought specifically by this amazing Nautilus article about plant cognition. It starts here, but there are multiple relevant bits that I won’t paste all in here for time’s sake:
“We need to get away from thinking of ourselves as machines,” Barrett says. “That metaphor is getting in the way of understanding living, wild cognition.”
Instead, Barrett and Calvo draw from a set of ideas referred to as “4E cognitive science,” an umbrella term for a bunch of theories that all happen to start with the letter “E.” Embodied, embedded, extended, and enactive cognition—what they have in common (besides “E”s) is a rejection of cognition as a purely brainbound affair. Calvo is also inspired by a fifth “E”: ecological psychology, a kindred spirit to the canonical four. It’s a theory of how we perceive without using internal representations.
I actually clipped more of it with additional commentary here. But I’ll dump one other bit of baggage before lift-off:
“The mistake was to think that cognition was in the head,” Calvo says. “It belongs to the relationship between the organism and its environment.”
I’ve seen a variety of different words for the mysterious fifth and sixth “E’s” in 4E/5E/6E cognition: ecological, emotional/affective, maybe “exaptive“? I’m not sure I understand how people use exaptive in that sense, and I prefer as the quasi-canonical base set of 6E the following:
- Embodied
- Embedded
- Enactive
- Extended
- Ecological
- Emotional/Affective
I had as a hypothesis that these 6E elements of cognition might have in them a good basis for ethics, and perhaps some clarity around the quagmire that is AI ethics.
I’m too lazy to do it all myself, so I had ChatGPT 4 help flesh out that argument, reproduced below:
- Embodied : The embodied aspect acknowledges that our physical bodies and sensory experiences are integral to cognitive processes. Ethical considerations arise as individuals recognize and empathize with the physical and emotional states of others. This recognition can lead to moral values centered around physical well-being and respect for bodily autonomy.
- Embedded: Embedded cognition emphasizes the influence of social and cultural contexts on our thinking. Ethics emerge as individuals navigate and contribute to the societal norms and values of their communities. This context shapes notions of right and wrong, justice, and communal responsibilities.
- Enactive: Enactive cognition focuses on the dynamic interactions between an individual and their environment. This interaction leads to ethical considerations about how one’s actions affect the world and others. It underscores the responsibility of individuals to consider the consequences of their actions in the social and physical environment.
- Extended: Extended cognition involves external tools and systems in cognitive processes. Ethical issues arise regarding the use and impact of technology, information access, and the distribution of cognitive labor. Questions about data privacy, intellectual property, and the equitable distribution of cognitive enhancements are pertinent.
- Ecological: Ecological cognition integrates the broader biological and physical contexts in cognition. This perspective leads to ethical considerations about environmental stewardship, sustainability, and the human impact on ecosystems. It encourages a sense of responsibility for the planet and its diverse life forms.
- Emotional/Affective: Emotional or affective components of cognition highlight how emotions are integral to cognitive processes. Ethics arise from understanding and managing emotions, both personal and in others. This includes considerations of empathy, emotional intelligence, and the moral implications of actions that affect the emotional states of others.
Those are fine, but I think their one around “extended” is the least strong. I guess I’m simplifying extended to be less about technology systems, but as simply knowledge sharing. Communicating with others, storing information, sharing experience – or not.
The rest of what I see here seems to revolve around reciprocity strongly, namely that one is an embodied entity in a world where actions have consequences, and you can suffer the consequences of others actions as well. Namely, the Golden Rule applies.
One thing I don’t like in 6E that is squiggly is that in attempting (probably correctly) to be neutral in describing the phenomena at issue, it does not minimally require that the embodiment be biological. It can be robot, or a distributed sensor network, etc.
Analytically, that makes sense. But my bias as a biological being is towards biological life, and living biological intelligence, mind, and cognition. So for me, I would draw a line in the sand (recognizing that as tides shift, it may get washed away) at something maybe you could stupidly call 6E-BR. Or biologically-rooted as a baseline requirement to include in discussions of 6E cognition.
Generally the 5th E seems to be Ecological, which correctly suggests that whether you yourself are biologically embodied, or otherwise, you exist in a world of other entities which are. For my purposes then, I might personally insert as an update to the “Fifth E” that it become Eco-Biological, highlighting the links to and importance of living biological things in all cognition. Here’s an improved detail on that from CG:
Eco-Biological Cognition: This term emphasizes that cognition is not only shaped by the immediate ecological context (as in Ecological Cognition) but is also deeply rooted in biological processes unique to living organisms. It highlights that cognition in biological entities is intrinsically connected to their life processes, cellular and molecular biology, and the ecological niches they inhabit.
So, following on my train of thought as to the basis of ethics being tied up in 6E and reciprocity, I asked CG to identify the areas where current AI fails (this one didn’t include the BR criteria). It churned out:
- Embodied: AI lacks a physical body and sensory experiences in the human sense. While robots can have a form of embodiment, they do not experience the world as living organisms do. The subjective, bodily experience integral to human cognition is beyond AI’s capability.
- Embedded: AI systems can process and respond to cultural and social data but do not inherently understand or participate in human culture or society as living beings do. They lack genuine social intuition and the deeply ingrained understanding that comes from being a part of a culture or society.
- Enactive: AI can interact with the environment in a limited way, usually through pre-programmed responses or learned behaviors. However, they do not possess the dynamic, autonomous interaction with the environment characteristic of living beings. Their ability to enact cognition through physical interaction is limited.
- Extended: AI systems can integrate external tools and systems effectively. This is an area where AI excels, extending its computational abilities through networks, databases, and other digital infrastructures.
- Ecological: AI does not have an inherent understanding of ecological systems or a natural context. While it can process ecological data and assist in ecological modeling, it does not have an intrinsic connection to or understanding of natural ecosystems.
- Emotional/Affective: AI lacks genuine emotional experiences and the subjective quality of affective states. While AI can simulate emotional responses and recognize human emotions to some extent, it does not experience emotions itself.
So, it seems that AI systems, when attempting to apply ethical or “safety” decisions rely solely on extended cognition – e.g., information that was shared with them. They don’t have access to and cannot understand the bio-physical roots of ethics in the other 5Es, the E’s that are all of primary importance to us in our direct lived experience.
AIs then acting under supposed “ethical” guidelines are little more than tone-deaf low-dimensional simulations, like drawings in medieval bestiaries, clearly made by artists who had never once laid eyes on the animal in question and wouldn’t know them by sight if it bit them on the ether. We should not call that “ethics” at all, imho.
Best I could get ^ out of Dalle on short notice for that visual reference, but you get the idea…
Anyway, the re-centering on biologically rooted life as critical to an amended 6E cognition seems worth it to me as a humanist, who thinks there is “something special” about us as people that is worth protecting and uplifting, even if mostly we just fuck it all up IRL. We can’t expect to get anywhere either if we don’t first tend to cleaning our own house…
If its only in the “extended cognition” category that AIs potentially excel, we should not I think then get too caught up in believing that AI is anything close to “alive,” when it fails in so many fronts that are so tied up in definitions of life.
I had old CG run links between 6E cognition, its concept of mind, and definitions of life, and its results are good, if a bit repetitive of the key points (like much of this post, I suppose):
- Mind as a Dynamic, Interactive Process:
- 6E cognition views the mind not as a static entity but as a dynamic process deeply integrated with the body (embodied), environment (embedded), and actions (enactive).
- This perspective aligns with broader definitions of life, which often emphasize dynamic processes, adaptability, and interaction with the environment.
- Embodiment and Biological Basis of Mind and Life:
- Both mind and life are fundamentally connected to physical, biological processes. The embodied aspect of cognition emphasizes this connection, highlighting the role of bodily experiences in shaping mental processes.
- The definition of life typically includes biological processes and physical embodiment, reinforcing the link between the mind’s cognitive processes and the broader characteristics of living organisms.
- Embedded and Ecological Aspects: Interaction with Environment:
- The embedded and ecological aspects of 6E cognition focus on the mind’s interaction with social and environmental contexts.
- Life is also defined by its interaction with and adaptation to the environment. Living organisms, including humans, are not isolated but are part of larger ecological systems.
- Enactive and Extended Aspects: Mind Beyond the Brain:
- Enactive and extended cognition propose that the mind extends beyond the brain, incorporating interactions with the external world and use of tools.
- This perspective resonates with the concept of life as a system extending beyond individual organisms, involving interactions with and adaptations to external stimuli and tools.
- Emotional/Affective Cognition: Sentience and Conscious Experience:
- The emotional/affective aspect acknowledges the role of emotions in cognition, integral to the conscious experience.
- Sentience and the ability to experience emotions are often considered key aspects of life, particularly in the context of higher organisms.
- Life as a System of Adaptive, Interconnected Processes:
- The definition of life usually includes the ability to grow, reproduce, respond to stimuli, and adapt. These processes are echoed in the principles of 6E cognition, which view cognitive processes as adaptive, responsive, and interconnected with the physical and social environment.
So if AI fails on much of these criteria for “life-i-ness,” where one might wonder does this feeling of aliveness come from experientially when one is using it? It comes from us. It is our own aliveness that we are experiencing, reflected back to us through this tool that extends our cognition, and allows us to find, create, and share knowledge. It’s alive-ishness is not encompassed in or enclosed in its GPUs or its programming – it’s enclosed in ours. And ours is enclosed in ecology and encased in biology, and that is something beautiful, special, magical, important, and worthy always of uplift and protection over and above the passing needs of our tools and toolsellers.
Digression upon digression, so sue me. But another avenue I got down when trying to hammer out the details about biological rootedness, what that CG gave me some other concepts which are also useful for exploring a multi-level cognition that goes beyone the 6E’s, and includes potentially:
- Physically Situated Cognition: Instead of “embodied”, use the term “physically situated” to emphasize that cognition is influenced by the physical form and capabilities of an entity, whether biological or artificial. This term can encompass both living organisms and AI systems or robots, recognizing that their physical form – body or hardware – plays a crucial role in how they process information and interact with the world.
- Materially Influenced Cognition: Another alternative could be “materially influenced cognition”, which underscores that the material makeup (biological or synthetic) of the entity influences its cognitive processes.
- Form-Dependent Cognition: This term would highlight that cognition depends on the form of the entity, whether it is a biological organism or a machine. It acknowledges that the structure, capabilities, and limitations of the physical form, whether made of flesh or metal, shape the cognitive processes.
- Biologically Rooted Cognition: This term emphasizes that cognition is intrinsically linked to biological processes and structures. It highlights that the cognitive capabilities and experiences are fundamentally tied to living, organic systems. This term would specifically exclude non-biological systems, focusing on the unique aspects of cognition that arise from biological life forms.
- Evolutionarily Shaped Cognition: This term reflects the idea that cognitive processes are not just embodied in a physical form, but are also the result of evolutionary processes specific to biological organisms. It acknowledges that cognition in living beings has been shaped by natural selection, adapting to environmental challenges and opportunities over vast timescales. This aspect would explicitly recognize the role of biological evolution in forming the cognitive abilities and predispositions of living organisms, distinguishing them from artificial systems like robots or AI.
After that we finally landed on Eco-Biological as more encompassing, but I find interesting the distinction here between Biologically Rooted & Evolutionarily Shaped, especially if we start getting into blending machines and biological components. If you grow meat in a lab, it’s biologically rooted, but is it evolutionarily shaped? Perhaps in its genetic antecedents, but depending on the chimera’s current shape, we might be far from anything resembling ‘Natural Selection’ at this point. Or, similarly, and AI system might have evolutionary aspects, but not be necessarily biologically-rooted, as Butler warned about in the 1800s.
Anyway, I think those are all the bases I wanted to cover: Mind extends beyond the physical brain. Humans are not machines. Let’s prioritize living biology and participatory ecology over flat lifeless definitions and conceptions of technology and existence that would have us all become like machines ourselves or cogs in someone else’s.
Leave a Reply
You must be logged in to post a comment.