Social sciences use “positive” and “negative” in mildly annoying ways in that those terms have quite clear meanings in everyday language but here are used in positive descriptive rather than normative or judging ways.
Feedback Loops
Positive (i.e. self-reinforcing or explosive) feedback loops: More input leads to more input afterwards. E.g. having more money makes it easier to make even more money.
Negative (i.e. self-stabilizing or self-balancing) feedback loops: More input leads to less input afterwards. E.g. in Mario Kart if you’re far behind you get better items, but once your placement improves the pick-ups you get decreasingly useful; better placement leads to worse items.
Reinforcement
- Positive (additive) reinforcement: add good stimulus → behavior increases
- Positive (additive) punishment: add bad stimulus → behavior decreases
- Negative (subtractive) reinforcement: remove bad stimulus → behavior increases
- Negative (subtractive) punishment: remove good stimulus → behavior decreases
i.e. you want to get additive reinforcement/reward or subtractive punishments and don’t want additive punishments or subtractive reinforcement/rewards.