In order to begin looking at Artificial Intelligence, we need to understand some of the fundamental principles on how it works. To do that, we take a quick detour into a little bit of behavioural psychology, philosophy and even a tiny bit of chemistry. With this in mind, we can then begin to look at how artificial intelligence and games adopt many of these principles.
Note: I’m neither a psychologist nor a philsopher or even a chemist. So this is a rather simple overview of some of the relevant topics.
Why Are You Reading This?
It’s a rather legitimate question given the subject at hand. Why are you here? What drove you to visit this webpage? Is it perhaps driven by curiosity? Or a legitimate urge to learn something new? Regardless, you had an agenda. You wanted to get something out of the experience. Now it might not be immediately obvious to you, but that is why you’re reading this.
At the end of the day, it’s because you want some sort of reward. Though in this case I don’t mean a monetary reward or a pat in the back. Rewards can come in all shapes and sizes: either the result of some extrinsic motivation (where someone offers you a reward for completing a given task), or an intrinsic motivation that is governed by yourself. These motivations are what push us, drive us as people to complete tasks; knowing that something good will come from it all (Ryan & Deci, 2000).
For humans, ‘rewards’ and the positive or negative feelings they elicit are the result of neurochemistry: given that the hormone dopamine is released into the ‘reward system’ of the brain as a result of committing certain actions. While still largely open to debate among neuroscientists, it is argued that the dopamine levels reinforces in our mind whether decisions we make are good or bad. (Arias-Carrión and Pöppel, 2007).
How these rewards manifest in the real world can vary. Here’s just some examples that come to mind.
- Your Choice of Lunch: You picked something that you will enjoy. You’re motivated to go to that one place you really like because they make the best slice of pizza. Interestingly, dieting can be called into question here, since the ‘reward’ is a healthy option that deep down you might not be happy with, but we’ll come back to that.
- Earning Money: Perhaps you work a full-time job, or you’re part-time while at college/university. Either way, you’re doing the job in order to make money. Heck, you might be working multiple jobs to make ends meet thanks to your country’s economy! It’s an explicit reward that is made clear to you by the contract of your employment. It’s also, for many people, the only reason you even bother doing the job. Of course, you can always earn money by other means.
- Love: That guy or girl you really like is pretty crazy about you too. You do things for them as an act of kindness, which helps to reinforce that relationship and the love you have for one another.
- Earn Respect: Perhaps you’re working hard for that promotion, or you really like that class you’re taking in school. You want to do well and earn the respect of your peers or perhaps the class tutor. Of course each of these also has an explicit reward that can equally justify your motivations.
Long Term & Short Term Rewards
- Short Term: Doing something which you will gain from almost immediately. Eating that slice of pizza you ordered or impulse purchases while shopping in town are prime examples of this. Interestingly, these have the potential to be detrimental to our wellbeing, given we can end up broke or unhealthy.
- Long Term: These often come from a long period of activity where we know that at the end of the day, all will work out in our favour. Gaining an education over three, four, five, possibly ten years of study at university is a prime example. Hitting the gym and getting the body you want is another long-term goal that often relies on us to fight the temptation of explicit, short-term reward (i.e. pizza).
The impact of rewards all boils down to how they govern decision making. People are influenced by the reward that is available to them and it dictates what actions we take. Our decision making can be condensed to these factors:
- Beliefs – Our knowledge of how the world works, which may not necessarily be correct, given it is our understanding of facts and rules that we have accrued over time. We can use these basic facts and rules and through inference learn new knowledge. We believe that taking certain actions will yield particular rewards.
- Desires – Our beliefs give motivation for desires we wish to accomplish. Our desires manifest goals that we will pursue.
- Intentions – Our intentions drive our actions to satisfy certain goals.
If we consider Ryu of the Street Fighter series, his goal is to become the strongest warrior he can be. He has concluded that the only way to do that is to continue to train and fight. It’s a long-term reward, which will only be gained after pursuing this goal to the best of his ability. As a result, he continues to participate in the Street Fighter tournament, given he believes these actions will move him one step closer to his goal.
Types of Decisions
How decisions are made can vary and the human mind (and body) have a part to play in that respect. We can break this down into two categories.
In some instances, we make decisions through deliberation: where we try to find a solution that satisfies a particular goal we wish to attain. Typically this results in a plan, where we have a number of actions that ensure we ultimately satisfy our goal. Why we make these decisions in the way we do has been the subject of discussion and debate for thousands of years in philosophy and is commonly referred to as ‘action theory‘ (White, 1968).
While rather tangential, the Joker in The Dark Knight points out that we only really panic when things don’t go according to ‘the plan’. That’s arguably because the plan made sense to us and we believed it was the correct course of action to take. When things don’t go to plan, we have to figure out what to do instead. Now the Joker’s actions and comments are more of a commentary about western society and the moral and judicial systems we have built around us. As a democratic society, we deliberate on goals we wish to achieve and are aware of the consequences that will come. It is only when actions occur that we did not expect, with consequences we did not already agree (or submit) to, that people begin to panic.
Conversely, we are capable of actions that border on involuntary, where the mind does not take time to deliberate, but reacts. In these instances, we know from experience what action is the best to take at this juncture. Reflex actions made when we burn our hand or dodging something thrown at your head are governed by the brains ability to quickly recognise and react to stimuli your senses have acknowledged (Purves et al., 2011).
It is this notion that makes Spider-Man’s ‘spider sense’ such a great super power: given it can detect threats and allow him to commit reflex actions without neither the need to think about that action nor to have witnessed the event that triggered his action.
This allows Spider-Man to detect potential threats to himself and others before they actually happen. But it does mean he does not know what will cause them until it happens.
So you may be wondering what does any of this have to do either with Artificial Intelligence or with games? Well video games are heavily reliant upon this: given it helps us not only to understand the games we play, but drives us to continue to play them. In addition, our discussion of knowledge, goals, actions and rewards are the pillars of foundational AI theory.
In Part 2, we discuss how these principles dictate how we play and make video games.
Arias-Carrión O, Pöppel E (2007). “Dopamine, learning and reward-seeking behavior”. Act Neurobiol Exp 67 (4): 481–488.
Ryan, R. M., & Deci, E. L. (2000). “Intrinsic and extrinsic motivations: Classic definitions and new directions.” Contemporary educational psychology, 25(1), 54-67.
Purves, D., Augustine, G.J., Fitzpatrick, D., Hall, W.C., LaMantia, A.S., White, L.E. (2011) “Neuroscience”, Sinauer Associates, Inc., Fifth Edition. ISBN 10: 0878936955
White, A.R. (ed.) (1968) “The Philosophy of Action” Oxford University Press.