'Games as learning' is a popular lens through which to see videogame play, and one with many advocates, not least of which being Raph Koster. His book, A Theory of Fun for Game Design, is a wonderful defence of the videogame industry built around the 'games as learning' premise. But when we use this viewpoint, what are we missing?

I have a number of issues with the 'games as learning' paradigm, none of which invalidate the merits of this point of view. My principle complaint is that, excluding reflexes, all behaviour is learned; almost all experiences in life are learning experiences. All creatures with a sufficiently complex brain learn all the time from what's going around them. Thus 'games as learning' doesn't track directly – if we are always learning, what's the difference between watching a movie and playing a game? The reason that games become more associated with learning is because games teach specific skills, whereas the kind of learning that goes on with a movie (or some other activity)is considered more passive. Thus rather than a strict case of 'games as learning' we end up in the position that games are simply efficient tools for learning.

But here we run into a serious issue: games teach skills by offering challenges and assigning tasks, which the player must learn how to complete. Clear a level, beat a boss, collect items in an area, and so forth. The player then receives a reward, and goes onto new challenges and tasks. The reward could be any number of different things, but biologically it all comes down to triggering the reward system in the brain, which I have already written about at length. The connection between this reward system and learning is well established – in fact, most professional animal trainers (such as those who train the orca's at SeaWorld) use the principles of operant conditioning in their work.

So now we are facing a parallel between what a videogame does to its player, and what a trainer does to an animal – with one big difference. The trainer is seeking to elicit specific behaviours – sit, wait and so forth when training a dog, for instance. The videogame has not been designed on this paradigm (the 'games as learning' model reflects a perception of videogames, not a prevailing attitude in how to make them), and indeed most videogames are merely slightly reworked versions of previous videogames with no intentions beyond being “better than other games”.

What do videogames teach? Well, among other things they teach players to smash boxes to look for items, to search the dead-end first, and in many cases to save often. Most of these skills do not transfer to real life in any useful way, although of course, there are exceptions (anyone who used Microsoft Word in the 1990s, for instance, will have benefited from learning to 'save often'). It is the case that learning to play most games does exercise the decision centre of the brain (the orbito-frontal cortex), and thus can help people with their problem-solving skills, although this was perhaps far more true in the early days of videogames than it is now. (I'm at a loss to suggest what is learned by a player of Wii Sports or, for that matter, Call of Duty: Modern Warfare, that they can apply outside of a videogame context, although I'm sure people will have interesting suggestions!).

As a consequence, I find the 'games as learning' paradigm flawed. I do not deny the potential for games as tools for learning, although the videogames industry has not embraced this very well and the Serious Games movement seems to have problems delivering on its full potential right now. Since 'games as learning' equates to 'games as tools for operant conditioning' I wonder if it wouldn't be more prudent to consider 'games as rewards' (as with John Hopson's Behavioural Game Design model) – both learning and rewards being aspects of the operant conditioning process i.e. the mechanism by which behaviours are learned.

The problem with 'games as learning' in this context is that it draws the individual's attention to an outcome (learning) and risks misleading people as to what actually goes on while a player is enjoying a game. To make his theory of fun track, Koster has to eliminate all forms of visceral fun, since these are not based upon learning, despite the fact that, yes, rollercosters are fun even though they do not involve learning.

Conversely, 'games as rewards' does not fall prey to this limitation so easily – from this perspective, visceral enjoyment is intrinsically rewarding, and thus can be sought for its own sake. Furthermore, 'games as rewards' has specific lessons for game designers as to the mechanisms they can use for structuring those rewards. Admittedly, one can easily invert this and be back in the 'games as learning' perspective, but those simple, visceral rewards – such as the thrill of high speed and the delight some players derive from violence – will never comfortably fit into this model.

My suspicion, perhaps unfair, is that for those of us who perform well in the education system learning is intrinsically rewarding (or autotelic, as Csikszentmihalyi says) – and this, I believe, applies to everyone who advances the 'games as learning' perspective and pretty much all game designers. Because of this inherent bias, 'games as learning' looks like a stronger model than it perhaps should. I feel 'games as rewards' is a required counterpoint and, from the point of view of game design, potentially more useful, while acknowledging that (like the ridiculous narratology vs ludology debate) we're really talking about two sides of the same coin.