But why is minimising surprise the very same as living longest? — Banno

Minimizing free energy means living a non-agitated, calm, life, with a minimum of energy use, in a silent dark room with optimal temperature and nothing happening. Eating natural stuff, sleeping a lot, without thinking. I think your dreams get pretty wild though. Your brain will not develop (as torturing rats has shown). Eventough dreams try to accomplish this. You theoretically will live longer, but probably die of boredom or go mad.

Minimizing free energy means living a non-agitated, calm, life, with a minimum of energy use, in a silent dark room with optimal temperature and nothing happening. — Cartuna

So lockdown is a boon for all!

No, I think the discussion in the article explains why the formula you suggest is not quite right.

So you are right that "your brain will not develop"; but a brain that does not develop would itself be a surprise; and hence enough to get one outside. At least, that seems to be roughly the article's argument.

So you are right that "your brain will not develop"; but a brain that does not develop would itself be a surprise — Banno

I see what you mean. So to evade the thrill of that surprise you escape the dark room to meet other surprises? On return to the dark room though, the non-evolving won't be a surprise no more though. Maybe for your unconsciousness, which might react again by sending you out, or making you dream to evade the surprise of not evolving.

Evolution? The folk expression might be that we avoid the dark room in order to avoid boredom; that is, we expect a certain level of sensory input, and find too little such stimulation too much of a surprise; hence we seek more stimulation.

Is this another free logic example? We don't minimise surprise as I understand. The brain rewards surprises in learning. Young children in particular pay most attention to novelty.

But why is minimising surprise the very same as living longest? — Banno

Friston states where the flaw in this lies...

The “reduce-surprise/live-longer” hypothesis seems to contain some remnant of the assumption that surprise is somewhere “out there,” a real, objective and measurable property of the world. In fact, it is subjective and relative to the interpretation applied by the agent, i.e., always “in the eye of the beholder.”

The whole "darkened room" angle only has legs for the naive realist.

Bayesian mechanics is about the pragmatic problem of taming the "blooming, buzzing confusion of reality" - the sensorium of the newborn - and turning it into a useful model of "the world as it would be with us in it". That is, a semiotic unwelt, or system of interpretance.

So whether we are talking of semiosis at the level of genes, neurons, words or numbers, the game is the same. Biology is negentropy in command of entropy production. It is a state of far from equilibrium thermodynamics (ie: dissipative structure) where some autonomous/emboddied self is in charge.

That is what Friston's Bayesian mechanics captures. Selfhood emerges to the degree the natural uncertainty of any dissipative environment can be made predictable and thus controllable.

It is not the world that must be made stable and boring. That is always going to be chokka with uncertainties. It is the ability to then forge a constant sense of intentional self within that blooming, buzzing confusion that is the task. It is selfhood that must be made stable and boring ... which is something most of us successfully achieve as we become creatures frozen into familiar routines and familiar concerns as the "way of life that defines us as being us, and not someone else". :razz:

Applying the minimization of free Gibbs energy to psychology is contrived. When free energy decreases entropy increases, but to form biological organisms in the first place the free energy has to increase. There is nothing that forbids the free energy from increasing during lifetime. There is no reason to expect that systems strive for minimizing free energy. Especially not biological systems, which take in energy to combat entropy reduction and preservation of ordered structure. Why should an animal strive for maintaining the status quo only, which is indeed achievable in dark rooms? Isn't there proof that this is not the case? A dark room might be a total surprise, totally unexpected, and will increase free energy. Luckily!

Is this another free logic example? — Kenosha Kid

Not that I noticed...?

"Surprise" here has a special meaning... "the difference between an organism’s predictions about its sensory inputs (embodied in its models of the world) and the sensations it actually encounters".

"Surprise" here has a special meaning... "the difference between an organism’s predictions about its sensory inputs (embodied in its models of the world) and the sensations it actually encounters". — Banno

Yes, same meaning I'm employing. We're quite surprise-a-philic if anything.

Indeed, free-energy looks like a piece of creative accounting, in the same way as potential energy needed to be introduced in order for the laws of thermodynamics to hold. But it works. So even though potential energy appears contrived, it's indispensable.

But why is minimising surprise the very same as living longest?

same meaning I'm employing. — Kenosha Kid

That doesn't appear to be the case. Children would expect novelty, and be surprised at its absence. Hence, they are curious.

Yes, same meaning I'm employing. We're quite surprise-a-philic if anything. — Kenosha Kid

In article:

One question that thus arises concerns the relation between all this talk about surprise minimization in the informational sense, and surprise minimization from the point of view of an intelligent agent. Perhaps it is too obvious to be worth stating, but we should also bear in mind that the two are remarkably distinct. One way to see this is to reflect that the state of the brain that most thoroughly – across all the time scales of adaptation – minimizes informational surprise may, at times, be a state that corresponds to a very surprising event or percept as far as the agent herself is concerned. For example, if I perceive a pink elephant in the middle of the room, that percept must itself be the one that – taking all those time scales (i.e., experience) into account – most effectively minimizes the long-term average of surprise about such data, conditioned on a model (again, in the inclusive sense of model). We should thus remind ourselves that even surprise relative to our best model can be tolerated, as evidenced by surprisingness to the conscious agent who may often – though not too often on pain of death – find herself in quite surprising and unexpected situations.

The affective state of surprise isn't what's intended by surprise minimisation. Worth keeping in mind it's a technical term in the underlying theory.

If life is the complexification of processes that lead to ever more elaborate organisms, given the right atmospheric conditions, then the first life to emerge would be quite "stupid" and have an incredibly poor model of the world.

On these conditions, the dark room would not matter, the organism would just "do stuff", to release energy and go back to oblivion. This tendency of doing stuff for the sake of it would carry on to most species as a residue of very primitive impulses. A dark room provides no stimulus and no potential reward. It would be safe, until the organism starves.

Better to release that energy doing something, while trying to be mindful of minimal safety precautions. Everything is going to perish anyway so, there's not much risk. In the grand scheme of things, living a few more months or years, is nothing compared to experiencing many things in a short time frame.

Yep.

If biological systems, including ourselves, act so as to minimise surprise — Banno

Novelty is rewarded with dopamine in just about all animals.

Yes, that seems to be about it.

But again, why should minimising surprise be the very same as living longest? It seems that in the account given in the article, it's not just that minimising surprise leads to long life, but that they mean the same thing.

I'm missing something. Can some explain it?

See my reply to @Kenosha Kid. and @fdrake's, too.

The Theorist in the article (the link that Banno provided) is making it more complicated than necessary. Pay attention to the Physicist's reply to the Theorist here:

Physicist: From the point of view of the free-energy formulation there is no need to recourse to any other principles. Of course, one might find that one’s favorite principle emerges from a particular application of the free-energy principle; however, the whole point of the free-energy principle is to unify all adaptive autopoietic and self-organizing behavior under one simple imperative; avoid surprises and you will last longer.

And in the beginning of the article, the Philosopher has also laid down the point of free-energy:

Philosopher: The “free-energy principle” (see e.g., Friston and Stephan, 2007; Friston, 2010a) suggests that all biological systems are driven to minimize an information-theoretic (not thermodynamic, though the two are mathematically close) quantity known as “free energy.” Free energy, as here defined, bounds surprise, conceived as the difference between an organism’s predictions about its sensory inputs (embodied in its models of the world) and the sensations it actually encounters. In this discussion, surprise is used explicitly as a measure of improbability from information theory. This is also known as surprisal or self information.

This is the principle of adaptation. The better one can interpret the sensory input of their environment, the better they can adapt for survival or whatever is needed to live as long as possible. The information theory deals with what's expected -- less surprise, or uncertainty, or unexpected, better prediction.

Children would expect novelty, and be surprised at its absence. Hence, they are curious. — Banno

I don't know about that. I know children are powerfully rewarded when they experience novelty as long as they feel safe.

It's why falling in love is so intoxicating. That's all dopamine.

It's not clear based on what they say. Perhaps we should distinguish surprise from stimulus and think of it as a kind of continuum. An organism would want to avoid surprises, meaning life-threatening situations, while seeking stimulus, a way to channel and then release energy.

Unless one stipulates that surprises must be avoided for survival. Then perhaps surprise is a bad term and we'd need a new one, such as "threat".

Research with mice shows it's not just stimulus that produces a hormonal reward. It's surprise.

The affective state of surprise isn't what's intended by surprise minimisation. Worth keeping in mind it's a technical term in the underlying theory. — fdrake

Going back to the OP:

If biological systems, including ourselves, act so as to minimise surprise, then why don't we crawl into a dark room and stay there? — Banno

With our without pink elephants and interesting edge cases, it isn't true that we minimise surprise in the way Banno is suggesting. In particular:

We should thus remind ourselves that even surprise relative to our best model can be tolerated, as evidenced by surprisingness to the conscious agent who may often – though not too often on pain of death – find herself in quite surprising and unexpected situations.

is not at all (ahem) surprising, since there is no general rule of aversion to surprise, nor is one needed to explain why people don't run at spikes, off cliffs, or into animal enclosures.

I know children are powerfully rewarded when they experience novelty as long as they feel safe. — frank

:up: And they need to be taught safety. Most young children are pretty fearless about climbing out of their cots (or up to the cookie jar). Certainly there are self-preservation instincts at work as well.

Conditions for those experiments are a bit suspect. They put a mouse in a cage with a lever containing some kind of drug. With nothing else to do, bored to exasperation, they'll pull the lever: it could be good or it could be bad. It's better than endless staying still.

When they recreate these experiments in a social setting, with many mice containing other stimulating things like small mounds and wheels and the like, barely any mouse opts for the lever. Some do, but a very small amount, which sounds correct given a sample group.

Conditions for those experiments are a bit suspect. They put a mouse in a cage with a lever containing some kind of drug. With nothing else to do, bored to exasperation, they'll pull the lever — Manuel

I was thinking about the one where they dropped food in a pattern for a while and then started randomly breaking the pattern. The result was a dopamine hit and accompanying frenzied behavior.

I can quote it from a book if you're interested in the details.

Why is that surprising? It's a novelty and they still get food.

I think we're assuming surprises must be negative or have negative connotations.

I don't doubt that experiment, but I don't see what is revealing about it. What the alternative, because the mice get food in a different pattern, they're just not going to eat?

What the alternative, because the mice get food in a different pattern, they're just not going to eat? — Manuel

They got used to a pattern and then they were surprised by a random deviation from the pattern.

The result was a dopamine release, which speeds up the heart and creates a sense of well being. Dopamine is the reason animals explore around in their environment. It's the reason they don't just hide in the dark.

It makes sense given those circumstances.

Please be aware that "surprise" is being used as a technical term.

Further, I am no advocating the argument in the linked paper, but raising it as an interesting topic.

Most of the posts here seem therefore off topic.

The article specifies that events with high surprisal are surprising in the common sense of the word.

Your posts are not the least bit surprising, tho.

The article specifies that events with high surprisal are surprising in the common sense of the word. — frank

Really? Where, exactly?

We have

In this discussion, surprise is used explicitly as a measure of improbability from information theory.

and admittedly,

Although the psychological notion of surprise is distinct, events with high surprisal are generally surprising.

And the connection to free energy

With this simple addition, we are now in a position to consider behavior and self organization; however, the same basic principle remains – namely, minimizing free energy or surprise.

Show me where it says it is used in the "common sense' sense.