Part 1: When the Mind Feels Nothing. What It Means for the Machines We Build
When I first wrote a slightly (or not so slightly) chaotic document exploring ideas linking Damasio and AI; neuroscience and computer science – at the time, I didn’t yet know that I would publish it, just as I didn’t know how important writing would become to me.
What began as a private reflection evolved into deeper and all-consuming research to understand emotion and logic in AI and in ourselves.
As AI continues to shape our choices and our relationships, understanding what makes human intelligence whole has never been more important.
This is a three-part series in which I explore how neuroscience explains why I see current AI as fundamentally incomplete and how we could build something better – and whether we even can.
About This Series
Part 1: When the Mind Feels Nothing. What It Means for the Machines We Build (You’re reading it right now!)
Elliot’s story and what it tells us about how the brain works. Why emotion isn’t the enemy of logic, but it’s infrastructure.
Part 2: Current AI’s Limitations
How today’s LLMs are essentially “logical Elliots” – brilliant at thinking but paralyzed where choosing requires feeling.
Part 3: Framework and Practical Applications
A technical breakdown of how we could build emotionally-aware AI systems – with concrete examples and ethical considerations.
Introduction
Usually, when the topic of intelligence comes up, it’s about knowledge: solving complex problems and chasing perfect answers. We rarely add caring to the equation.
And one day I found myself thinking that maybe the key to building better AI doesn’t lie in making it smarter (or rather, not only that), but in understanding why intelligence without emotion collapses. And I remembered the story that comes from a man who lost the ability to feel. And with it – the ability to live fully.
Meet Elliot – a patient studied by neuroscientist Antonio Damasio. After a brain tumor was removed, his IQ remained intact. He thought brilliantly, logically, and… became completely dysfunctional.
The case notes say Elliot couldn’t make decisions. He’d spend hours choosing between kinds of cereal, paralyzed by trivial choices. He lost his job. His relationships dissolved.
It’s very unsettling to realize that a person can lose the ability to feel but keep the brilliance of their mind. Just think about it for a second. How did it come to be that logic alone wasn’t enough to hold a human life together?
This case shattered the myth that emotion and reason are separate. It revealed something profound: emotion isn’t the enemy of logic, but its essential partner.
After all, each of us can get stuck between choices: two job offers, a relationship, or even just which restaurant to pick. Logic lists all the pros and cons, but what we actually need is to stop… and feel what to decide. Aren’t we? ;)
The Neuroscience: Emotion as Infrastructure
What Emotion Actually Is (And It’s Not What You Think)
We usually see emotions as a vague, subjective feeling that, honestly, sometimes just gets underfoot – something that clouds true, cold reason. We’re taught that logic is superior, and that emotions can’t be trusted.
But neuroscience says the opposite.
Let me start with a couple of definitions:
Emotion is the process of routing signals that connects what we think to how the body reacts. It isn’t separate from thinking – it is part of it. It’s signal, not noise. It’s infrastructure.
How does this differ from intellect?
I’d define it like this – especially given AI and how it’s integrating into our lives:
Intellect is a combined process in which several systems work together. It’s the ability to read all incoming data: what you see, what you feel physically and emotionally, and to connect that with memory, knowledge, experience, and the search for new information. All of it is used to create something new or solve problems.
As we can see, intellect isn’t separate from emotion at all. Intellect is an abstraction, and emotion is one of its key components.
This brings me back to the wonderful Jeff Hawkins and his book “On Intelligence” (Hawkins, 2004), and his formula: Intellect = Memory + Prediction. Wait – but he didn’t mention emotions at all. So is his formula wrong?
No. It’s not wrong. It’s incomplete. Because Hawkins had his own goal.
Hawkins wasn’t trying to build a full cognitive architecture. He focused on sensorimotor prediction, in particular on modeling cortical columns. How we predict patterns based on memory and incoming data.
And as he himself wrote: “I am not interested in building humans. I want to understand intelligence and build intelligent machines. Being human and being intelligent are separate matters.” – and this quote tells us he isn’t wrong. We’re simply after different goals, and I’m building on his book!
I, on the other hand – almost naively, some might say – am trying to capture every detail of the “mechanism.” And in it, the emotional component is present – the very one that turns prediction into action.
Where Logic Lives – and Where It Stops
Logic draws on several regions of the brain:
- Dorsolateral Prefrontal Cortex (dlPFC): the “chess grandmaster” – slow, controlled, symbolic reasoning
- Parietal Lobes: visual-spatial logic, geometry, coding
- Left Temporal Lobe: verbal logic, the structure of arguments
The breakthrough came when neuroscientists realized that decision-making requires integration across systems:
- Amygdala: emotional intensity, danger/safety signaling
- Insula: interoception – feeling what’s happening inside the body
- Anterior Cingulate Cortex (ACC): reads bodily signals and integrates them
- Ventromedial Prefrontal Cortex (vmPFC): this is the crucial region. Modern neuroscience shows the vmPFC isn’t a single uniform zone – it’s functionally heterogeneous, with different subregions handling emotion regulation, value representation, and social cognition (Wallis, 2007; Roy et al., 2012). Broadly, this region integrates cognitive reasoning with emotional signals and somatic (body-based) feedback. When damaged, it leads to exactly Elliot’s symptoms – intact logic with emotional-volitional paralysis (Damasio, 1994).
How a healthy brain makes a decision:
- The autonomic nervous system activates (heart rate changes, gut tension, sweat)
- The insula reads these bodily signals – this is called interoception, the brain’s map of your internal state (Craig, 2002). The anterior cingulate cortex then integrates these signals with emotional context.
- The vmPFC weighs all of this – bodily sensations, emotional significance, and logical considerations – to form value judgments (Hare et al., 2009; Tusche et al., 2010)
- You get a final, weighted decision: not just “what is,” but “what matters”
Modern research emphasizes that decisions are a dynamic network where emotion, cognition, and body signals continuously interact (Phelps et al., 2014). And here’s why it’s important: your brain literally needs to feel your gut tension or your racing heart to make choices that align with what your body knows.
Even our “purest logical thought” isn’t purely logical. It’s constantly being colored, prioritized, and modulated by bodily states.
Logic can tell you what’s consistent, but not what’s good, important, or meaningful. Value itself is not simply discovered; it is constructed, and affect is part of that construction (Hartley & Sokol-Hessner, 2018).
From the ability to spot regularities, apophenia can be born – itself the seeing of patterns where none exist. But it’s emotion that convinces us those coincidences matter. That’s how coincidence turns into a Sign.
Logic can propose many solutions, but not tell you which one is worth choosing.
Evidence from lesion studies points to the vmPFC’s causal role in integrating these signals, though researchers are still working out exactly which subregions control which functions (Noonan et al., 2010; Hunt et al., 2012). But the pattern is clear: when this integration breaks down, decision paralysis sets in. Everything feels equally important – or equally trivial.
Logic, Emotion, and Body Sit Down to Negotiate
Decision-making isn’t a battle between logic and emotion. It’s not even a partnership. It’s a negotiation – between logic, emotion, and the body.
To illustrate, let me give a real example from my own life:
I wake up at 6:39 AM after a sleepless night (read… I never actually went to bed ;) ). I want coffee, but I feel fear in my stomach. Logic says: “You need to sleep. Coffee will only make it worse.” But another part wants coffee anyway. And a third suggests: “Write instead. This exploration feels exciting.”
Here’s what was actually happening:
- Coffee desire → Impulse (initial action bias)
- Sensation in the stomach → Somatic check (fear signal)
- It’s 6:39 AM → Rationality kicks in (time awareness)
- I haven’t slept → Rational fact (state awareness)
- I want coffee anyway → Emotional feedback loop
- I’ll write instead, it feels exciting → Reframing through synthesis
- I made a decision → Commitment and emotional closure
But this isn’t a linear process! Rather, it’s like three people talking everything over among themselves until they align. Logic. Emotion. Body. They’re not in conflict – they’re negotiating. They argue until they reach alignment. Not agreement, not “truth” – just alignment. A shared “yes.”
I want to note that “arguing” here isn’t the modern kind of quarreling it so often becomes. In Ancient Greece – the world of Socrates, Plato, and later Aristotle – there was this thing called dialectic: above all, a method of inquiry through questions, answers, objections, and testing what follows. And it didn’t necessarily end in agreement. Rather, in alignment: “We’ve established that our original answer is wrong, but there’s no final right answer yet – and we’re okay with that.”
And this is exactly how I picture our inner process.
This is a diagram of the whole loop – an attempt to visualize it from the body’s first signal, through the vmPFC weighing everything, to the decision that gets stored and shapes the next one:
graph TD AUT["Autonomic Activation<br/>heart rate, gut, sweat<br/>via brainstem + hypothalamus"] INS["Insula: Body Map<br/>interoceptive signals"] ACC["ACC: Integration<br/>& motivation context"] LOG["dlPFC: Logic<br/>options & patterns"] AMY["Amygdala: Intensity<br/>danger/safety"] VAL["vmPFC: Value Weight<br/>bodily + emotion + logic"] DEC["Decision: What Matters"] SOM["Somatic Marker<br/>body memory pattern<br/>stored via vmPFC-insula connection"] FUT["Future Reactivation<br/>feeling shortcut"] AUT --> INS INS --> ACC LOG --> VAL AMY --> VAL ACC --> VAL VAL --> DEC DEC --> SOM SOM --> FUT FUT -.->|Reactivation Loop| INS classDef logic fill:#d2e3c6,stroke:#758879,stroke-width:1.5px,color:#2d2d30 classDef emotion fill:#e3c6d2,stroke:#8b7a87,stroke-width:1.5px,color:#2d2d30 classDef body fill:#ddc6e3,stroke:#7e7a8b,stroke-width:1.5px,color:#2d2d30 classDef integrate fill:#5a5570,stroke:#3f3b52,stroke-width:1.5px,color:#f5f5f0 classDef decision fill:#c5e0cb,stroke:#758879,stroke-width:1.5px,color:#2d2d30 classDef memory fill:#e3d7c6,stroke:#8b7e7a,stroke-width:1.5px,color:#2d2d30 class LOG logic class AMY emotion class AUT,INS,ACC body class VAL integrate class DEC decision class SOM,FUT memory
Somatic Markers: The Body That Remembers
When we make decisions and live through their consequences, the brain ties part of that experience to emotional and bodily states. This memory is stored not just cognitively but emotionally, as a pattern of bodily activation (tension, nausea, excitement, warmth).
The next time you face something similar, your brain reactivates that feeling even before you can consciously think about it.
These are what Damasio called somatic markers – emotional cues that act like internal compasses, a concept he developed from studying cases like Elliot’s (Damasio, 1994).
We see a similar principle in emotional learning. For example, a girl eats ice cream while living through a traumatic event. Even years later, she might avoid ice cream – not because she remembers the trauma cognitively, but because her body remembers. The feeling gets tied to the experience. And now, ice cream = fear. This is a well-known phenomenon in therapy. It’s not deliberate reasoning. It may not even be conscious recollection. It’s an affective association that can be felt before it is understood.
Music works the same way. Sometimes when I hear a song, the scene doesn’t come first. First I feel it – instantly – and only then do I recall the situation (sometimes it takes me a very long time to recall). The whole emotional landscape floods back. It’s not a snapshot – it’s more of an echo in the body.
And this system can backfire. Sometimes we feel safe in situations that are actually unsafe. Why? Because if a person never learned what safety feels like, the unfamiliar – even if it’s objectively safe – triggers fear. And the familiar – even if it’s dangerous – feels “right.” Not because it is, but because it matches what our body knows.
Emotion isn’t just an influence. It’s infrastructure: it helps us decide and filter. But sometimes it becomes a trap – if our inner compass broke somewhere on life’s journey.
Let’s talk about interoceptive loops.
The brain doesn’t just read the body’s state once and make a decision. It receives and integrates signals from inside the body, predicts its state, and helps regulate it. The body changes and sends new signals, and the brain updates its model.
Like a continuous loop: the body shapes appraisal, appraisal shapes action and regulation, and the body’s new state shapes appraisal again.
That’s why a decision is never truly final.
When the Systems Don’t Align
When all three systems – logic, emotion, and body – are aligned, decisions flow easily. But what if they don’t?
That’s when people get stuck and lost. And there you are, lying awake the whole night, naively weighing options that seem equivalent and equally impossible. We’ve all been there – and now we know why it’s so hard.
Sometimes a decision has to be made, but none of the options feels right. The systems can’t align; no resonance is found. So the brain loops – going over the options again and again, trying to reduce the internal conflict and bring the system into a more coherent state.
If there’s no resolution, no good options – just the weight of having to choose – that’s why it hurts. That’s why it loops.
Interestingly, when I first published this article, I was living through exactly what I was writing about. I couldn’t figure out what to do in the circumstances I found myself in: logic, emotion, and body didn’t line up. It felt like I was writing from inside the very misalignment I was trying to describe.
And that’s exactly why we can’t know what we’d do in situations we’ve never been in.
Because it’s not just logic that decides. It’s emotion + memory + body + context. We might think we’d never lie, or that we’d always follow our ethics, or die for someone we love.
But until all three voices speak, we’re just guessing – believing in the best in ourselves, but we truly won’t know until the moment happens.
One of the mechanisms behind this not-knowing is called the “Hot-Cold Empathy Gap” (Loewenstein, 2005): while in one emotional and physiological state, it’s hard to accurately imagine how we’ll feel and act in another. It’s a well-studied cognitive bias. And the problem runs even deeper when it comes to circumstances we’ve never lived through.
And this brings us to empathy. Maybe empathy isn’t about understanding how another person feels. Maybe it’s about recognizing that sometimes we simply can’t understand. Because we weren’t there. Because our system hasn’t processed that exact configuration of fear, grief, shame, or love.
So maybe real empathy isn’t saying, “I get it.” Maybe it’s saying, “I know I might never fully get it. But I see that you do. And I believe you.”
And that’s enough.
A Crucial Distinction Before We Bridge to AI
When we explore computational analogues for emotion-like processing in Parts 2 and 3, we must be clear: these are behavioral simulations, not biological replications. Machines won’t “feel” emotions the way humans do. They won’t have somatic markers, interoceptive loops, or the subjective experience of fear, joy, or dread.
What we’re exploring is whether computational systems can simulate the patterns of emotional weighting and value-signaling that make human decision-making possible.
Elliot’s paralysis came from a breakdown in the biological mechanisms of emotional valuation and value-signal integration. AI’s paralysis may come from the absence of their computational analogues. Different mechanisms, potentially similar outcome.
A brief technical note: what I’ve called “emotion” throughout this piece encompasses overlapping processes – emotion (reactive value-signaling), motivation (goal-directed allocation), and affect (valenced states). The neuroscience I’ve drawn on treats these as integrated rather than modular: not three separate agents “negotiating,” but one hierarchical system (predictive coding, active inference) where valuation happens at every level. Exploring these analogues – somatic marker equivalents via reinforcement learning, Bayesian priors, or affective computing models – we’ll distinguish behavioral simulation (pattern weighting) from intrinsic motivation (genuine valuing).
Not feeling. Not consciousness. Pattern recognition and routing.
Conclusion: Logic That Cannot Choose
Elliot’s story leaves us with a simple, unsettling truth: logic alone can think, but it cannot choose. It can model the world, but it cannot care about it.
Emotion isn’t a luxury – it’s the pulse that gives reason direction. The mind doesn’t work through separation but through connection: thought, feeling, and body speaking in the same language of meaning.
When that dialogue breaks – as it did for Elliot – intelligence becomes weightless. It can analyze everything, yet move toward nothing.
In Part 2, we’ll turn this mirror toward machines. We’ll see how today’s AIs, for all their brilliance, resemble Elliot – perfect logic suspended in emotional vacuum, able to calculate but unable to value.
And in Part 3, we’ll explore how to change that – how emotion-like architectures could help machines not replace human judgment, but support it, learning to weigh context, care, and consequence.
We can’t build great intelligence that understands us until we remember what it means to be us.
What do you think? Have you noticed how AI systems struggle with prioritization or decision-making? What would an emotionally-aware AI look like in your field?
Thank you for reading and thinking alongside me. I’m genuinely glad you’re here. This is Part 1 of 3 – a series exploring how neuroscience reveals why current AI is fundamentally incomplete and how we could build something better.
References
- Craig, A. D. (2002). “How do you feel? Interoception: the sense of the physiological condition of the body.” Nature Reviews Neuroscience, 3(8), 655–666.
- Damasio, A. R. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. New York: G.P. Putnam’s Sons.
- Hare, T. A., et al. (2009). “Self-control in decision-making involves modulation of the vmPFC valuation system.” Science, 324(5927), 646–648.
- Hartley, C. A., & Sokol-Hessner, P. (2018). “Affect is the foundation of value.” In A. S. Fox, R. C. Lapate, A. J. Shackman, & R. J. Davidson (Eds.), The Nature of Emotion. Oxford University Press.
- Hawkins, J. (2004). On Intelligence: How a New Understanding of the Brain Will Lead to the Creation of Truly Intelligent Machines. New York: Times Books.
- Hunt, L. T., et al. (2012). “Mechanisms underlying cortical activity during value-guided choice.” Nature Neuroscience, 15(3), 470–476.
- Loewenstein, G. (2005). “Hot-Cold Empathy Gaps and Medical Decision Making.” Health Psychology, 24(4S), S49-S56.
- Noonan, M. P., et al. (2010). “Separate value comparison and learning mechanisms in macaque medial and lateral orbitofrontal cortex.” PNAS, 107(47), 20547–20552.
- Phelps, E. A., Lempert, K. M., & Sokol-Hessner, P. (2014). “Emotion and Decision Making: Multiple Modulatory Neural Circuits.” Annual Review of Neuroscience, 37, 263–287.
- Roy, M., et al. (2012). “Ventromedial prefrontal-subcortical systems and the generation of affective meaning.” Trends in Cognitive Sciences, 16(3), 147–156.
- Tusche, A., et al. (2010). “Automatic processing of political preferences in the human brain.” NeuroImage, 49(1), 914–923.
- Wallis, J. D. (2007). “Orbitofrontal cortex and its contribution to decision-making.” Annual Review of Neuroscience, 30, 31–56.