“Dimensions” because n real values describe an n-dimensional space.
Semantic labels are a way to describe emotions but would not normally be used in conjunction with dimensional descriptions. Typically you would see a 2D arousal-valence emotional space or a 3D arousal-valence-dominance space. (Dominance is a relatively recent addition needed to distinguish some emotions that occupy the same region of arousal and valence: one such pair is anger and fear, both high-arousal low-valence.)
I would say yes. If we take ambivalent to mean “having mixed or contrary ideas about something or someone” like my desktop dictionary defines it, you can definitely feel mild ambivalence (where the two or more conflicting feelings are themselves mild), or intense ambivalence, where you feel strong, painful conflicting emotions about something.
In my original comment I assumed at minimum that you’d have a (probably discrete) type of emotion, and definitely intensity, and maybe valence (no need for polarity if you just add more types). However, it occurs to me now that the intuition supporting discrete types of emotion (because they may be founded in different physical implementations) would also support a many-dimensional continuous strength-of-activation dimension. That is, I see no evidence that there’s only one currently felt emotion.
I followed your search advice and I think I do understand what the cited valence/arousal (both real valued) classification is. Maybe those are the two most important factors, but I’m not impressed.
I’ll postpone thinking/research about how to categorize emotion in favor of more practically useful things. I’ll be interested once researchers have an implementation-level argument for why their abstract emotional state space explains what really happens in brains (a simpler model that doesn’t have any correspondence to brain-level stuff could still make good predictions, but I’m not going to search for that unless anybody has a specific recommendation).
I’ll be interested once researchers have an implementation-level argument for why their abstract emotional state space explains what really happens in brains
Easy: we have separate hardware for approach and avoidance behaviors, rather than a single linear “what’s the value of this” system. It’s easier to first evolve systems for avoiding bad things and approaching good things, than it is to develop a decision-making system that weighs pros and cons and decides which way to go. You can develop a disambiguation system after the first two systems are there, but it’d be hard to make from scratch.
(This, btw, is why I think utility expressed as a single number is lossy with respect to human values: when humans have both utilities and disutilities in a scenario, they usually experience conflict, not neutrality or indifference!)
This, btw, is why I think utility expressed as a single number is lossy with respect to human values: when humans have both utilities and disutilities in a scenario, they usually experience conflict, not neutrality or indifference!
That is a very insightful comment, I find. Let me ponder on that...
I expect then that some approach-related and avoidance-related ‘emotions’ can co-activate (although you’d expect some mutual inhibition circuits, perhaps in some cases it’s mediated only in deciding what concrete physical action to take).
“Dimensions” because n real values describe an n-dimensional space.
Semantic labels are a way to describe emotions but would not normally be used in conjunction with dimensional descriptions. Typically you would see a 2D arousal-valence emotional space or a 3D arousal-valence-dominance space. (Dominance is a relatively recent addition needed to distinguish some emotions that occupy the same region of arousal and valence: one such pair is anger and fear, both high-arousal low-valence.)
This is precisely what I”m not understanding: how is intensity of valence different from intensity of arousal.
In other words, can I feel intensely ambivalent? If so, then I see why they claim >1 dimension. If not, I don’t follow.
I would say yes. If we take ambivalent to mean “having mixed or contrary ideas about something or someone” like my desktop dictionary defines it, you can definitely feel mild ambivalence (where the two or more conflicting feelings are themselves mild), or intense ambivalence, where you feel strong, painful conflicting emotions about something.
Neutral valence, high arousal could be “surprise”. If sustained, it’s the proverbial “state of cat-like readiness”.
I recommend using Google Images (“arousal valence space”) to find some pictures, which I think would help your intuition along.
In my original comment I assumed at minimum that you’d have a (probably discrete) type of emotion, and definitely intensity, and maybe valence (no need for polarity if you just add more types). However, it occurs to me now that the intuition supporting discrete types of emotion (because they may be founded in different physical implementations) would also support a many-dimensional continuous strength-of-activation dimension. That is, I see no evidence that there’s only one currently felt emotion.
I followed your search advice and I think I do understand what the cited valence/arousal (both real valued) classification is. Maybe those are the two most important factors, but I’m not impressed.
I’ll postpone thinking/research about how to categorize emotion in favor of more practically useful things. I’ll be interested once researchers have an implementation-level argument for why their abstract emotional state space explains what really happens in brains (a simpler model that doesn’t have any correspondence to brain-level stuff could still make good predictions, but I’m not going to search for that unless anybody has a specific recommendation).
Easy: we have separate hardware for approach and avoidance behaviors, rather than a single linear “what’s the value of this” system. It’s easier to first evolve systems for avoiding bad things and approaching good things, than it is to develop a decision-making system that weighs pros and cons and decides which way to go. You can develop a disambiguation system after the first two systems are there, but it’d be hard to make from scratch.
(This, btw, is why I think utility expressed as a single number is lossy with respect to human values: when humans have both utilities and disutilities in a scenario, they usually experience conflict, not neutrality or indifference!)
That is a very insightful comment, I find. Let me ponder on that...
It does seem easy. Thanks.
I expect then that some approach-related and avoidance-related ‘emotions’ can co-activate (although you’d expect some mutual inhibition circuits, perhaps in some cases it’s mediated only in deciding what concrete physical action to take).