Ask HN: Are emotions algorithms for survival? | Hacker News Reader

Ask HN: Are emotions algorithms for survival?

2 points16961714b6 comments

Some scientists argue that emotions are algorithms for survival. Yuval Noah Harari in his Homo Deus writes that emotions are biochemical algorithms that are vital for the survival and reproduction of all mammals. Similarly, Dominik R. Bach and Peter Dayan identify principles according to which these algorithms are implemented in the brain and describe their approach by considering decision making in the face of proximal threat. [doi:10.1038/nrn.2017.35]

But let's say the organism is addicted to a substance or destructive behavior, and its brain's reward center keeps telling it that it feels good to repeat a self-destructive behavior.

Is this an error in the algorithm, since it may pursue the opposite of survival? Are emotions really algorithms for survival?

View on HN

"Are emotions algorithms for survival?"

The trouble humanity has always had while contemplating her own nature, lies in complexity providing more than one solution to similar problems. Individually, our cognitive strategies diverge for greater collective coverage. Genetic efficiencies, experiential refinement, and deliberate manipulation vary results.

By the insights I have gained by life long introspection, I offer that emotional system provides "smoothing", and "weighting".

The weighting may be considered a form of survival algorithm insofar as it primitively associates with the mind's facility for inductive reasoning (defaulting to attachments at birth, growing more complex through reinforcement throughout the lifetime.

As smoothing, the inherently abstract nature of information is grainy. Consider what may make one piece of information more important than another. Meaning derives from long complex (and branching) chains from root implications to perceived values. The emotional subsystem provides a smoothing between independently compartmentalized neural processes.

"By default" emotions may work one way, with the reinforcement of training they may work a different way, with some special kinds of development they may be either overridden or enhanced to spectacularly un-intuitive results.

It depends on what you mean by survival, also classifying them as 'algorithms' is incredibly lazy thinking, doing that confuses the basic issue because you start to attribute properties of algorithms to emotions etc.

A good read on it is 'Why we cooperate' https://mitpress.mit.edu/books/why-we-cooperate by Tomasello, he comes at it from an anthropological perspective, a quick example is 'guilt', guilt is an emotion that allows other members of our tribe to see that we're sorry and to forgive us, if the tribe doesn't forgive then you're exiled and die. So your survival in that context is dependent on emotional responses etc, but reproduction certainly isn't dependent on emotions.

Shame is a firmware interrupt, as described in Donald Nathanson's Shame and Pride, a book on Silvan Tomkins' Affect Theory.

Interesting example with 'guilt' as an emotion that may help to foster cooperation. This reminds me of an idea of evolution being treated as a thermodynamic process and 'punishment' as a crucial mechanism in the evolution of cooperation. The authors argue that "punishment acts like a magnetic field that leads to an 'alignment' between players, thus encouraging cooperation."[1]

[1] https://www.technologyreview.com/s/608139/new-model-of-evolu...

1) Emotions are an evolved heuristic algorithm and only should be judged on a statistical basis. 2) The algorithm evolved to serve the species, not the individual. 3) Evolved algorithms can't be expected to handle circumstances not present during their evolution e.g. hard drugs and electricity.

On point 3, compare how well a cat's algorithm for landing on its feet works in its evolutionary adapted environment to how well it works in 0-g: https://www.youtube.com/watch?v=O9XtK6R1QAk&feature=youtu.be

On point 2, I feel you should caveat that's only if group selection is a thing, which many don't think it is.

Reply on news.ycombinator.com