Based on recent thinking, I’ll be publishing an article in the Sandhurst Occasional Papers that sketches out a mechanistic approach to ethical behaviour. How good people do bad things, but getting beyond the oddly superficial answers people usually resort to.
The full draft of my paper is here, but since it’s 45-odd pages long and includes formulae, I reckon you probably want something a bit easier (and less long winded) to chew on.
Now, frankly, I don’t have the time to write an article as an explainer. I spent all week drafting the actual article. So I thought I’d see if NotebookLM could create a good podcast on the topic—try a slightly different form of content.
It’s pretty good! Good enough to work as content this week. And here’s a very brief (also AI helped) explainer for those who prefer to read. Should be a transcript around somewhere too.
1. Why another model?
Ethics—certainly military ethics—still suffers from two old habits.
Reason worship – We teach people to “think it through”. Slow down, take a breath. In practice the thinking part often works like a press-office: it justifies whatever the gut has already decided, and it does so worst when we are tired, rushed or frightened—usually times when ethics is most important.
Prescription without plumbing – It’s one thing to tell people what they should be, but few people talk about how we can do it. This is what behavioural science is all about, so why not fix that.
Unless we know how moral behaviour is generated, we’re not likely to be able to do anything about it.
2. The ETHIC Stack in one breath
My little ETHIC Stack is a sketch of the kind of thing I reckon would fix these problems. It’s a a five-layer mechanism that traces the chain from a flash of emotion to a public act:
E — Early, emotional circuitry
This is the neural level. Stimuli get rapidly, and pre-consciously “tagged” with a valence—a sense of goodness or badness. Urges to approach (protect/care) and avoid (threat/disgust). These things trigger arousal surges that act like neural cash, “purchasing” highly-trained automatic action scripts. All this gets passed up to the next level as an “intuition”.T — Thought-level schemas
This is the cognitive layer. We can imagine it like a gate-keeper—the intuition enters, and unless there’s a good reason, that intuition sails right on through the gate, and out into complex action (as opposed to the more simple action routines already kicked off at the E-level). Sometimes, we feel the need to “shut the gate” and think about things. Maybe we’re faced with a “right vs right” decision, instead of a “right vs wrong” decision, for example. Sadly, when we close the gate, it’s not often that we’ll think things through objectively. Usually this level works to rationalise the more convenient thing to ourselves.H — Habitat (the immediate situation)
This is a silly name, but it makes te acronym work. I just mean the current situation. Lighting, noise, time-pressure, sleep-loss, the physical layout of the world, even the order in which options appear on a form—all of these nudge which actions feel most obvious and affordable. Psychologists call them “affordances”: the environment quietly bids you to behave in certain ways, based on what aspects are most obvious. See a vertical pole on a lurching subway, and you’ll see it as an opportunity to grab it and stabilise yourself. If the subway is smooth, you’ll ignore it and see the seat instead.I — In-group dynamics
People don’t obey rules so much as identities. The action that wins the auction of possible actions in your head is the one the group you feel fused with expects. A single respected leader shaking his head, or a buddy saying “we’re better than this”, can change everything. Prestige signals, rumour mills, WhatsApp chats and gallows humour are the pipelines through which norms, pride and disgust spread.C — Cultural scaffolding
Doctrine, law and myth supply the priors: what counts as right, what counts as shameful. This is where norms are injected, and those norms depend on where and who you are. There isn’t much we can do about this, but without this layer, all the model is is an incomplete account of human behaviour, rather than a sketch of moral behaviour. Worth considering.
Intuitions appear at every level, informed by the cultural norm-injection. But each level implies different things for what will lead people to do unethical things.
4. A four-gesture drill
How do you use it? Well each level implies a different flag. If you’re feeling a lurch of adrenaline, or notice a half formed excuse, then you’re dealing with those early levels. If it’s a dark joke that makes you uncomfortable, or an unclear instruction, then it might be the more social levels.
Work out which level is loading the most moral risk (you’re angry; you’re in a thought spiral; you’re in a dodgy setting; there’s peer pressure), and then you can grab the cheapest handle that neutralises that level. For example:
Box-breathing or practice a new rehearsed script for E.
Say the doubt aloud for T, ideally to someone else.
Try to notice more than just the obvious thing in the Habitat.
Ask yourself what someone you respect might do for I.
Switch normative frameworks—remind people there’s more than one way to view the world for C.
And then you iterate. Did you drop the moral risk? If not, go again.
Outro
I’ll do this article properly at some point. But for this week, it’ll do.
If anything here seemed interesting, the full draft is here. I think I’m onto something. You can let me know.
Share this post