There is a disconnect when we think about probability and thought. To think that not only do we calculate the probability of events, but that we do it many times a day subconsciously can be quite suspect. Still, this idea makes sense. If we see that the ground is wet we immediately deduct that something made it wet. We then take the information of our environment to guess the cause. If the sun is shining there is a strong chance that it did not rain. If we see a sprinkler close by there is a strong chance that it could have gotten the ground wet. Even though we do not formally think P(sprinkler/wet) = (P(wet)*P(wet/sprinkler))/P(sprinkler) this formula does yield similar results. It seems that we do something close enough so that this model accurately reflects (or at least describes) some of our thought process. This is evident in today’s expert systems, some machine learning algorithms, and computer vision. What interest me more, however, is how we arrive at these prior probabilities, and how often we change them.

The calculation of some probabilities seem straight forward. If it is raining things will get wet would imply that P(wet/rain)=1. Yet, if something is blocking the rain (like a roof or a tarp) this is not true. The condition of wet remains the same, but P(wet) is either dynamically calculated or we do not calculate P(rain/wet) but instead P((rain/wet)/tarp). It seems to me that something other than just Bayes theorem is being calculated. Do we calculate heuristics?  Do we repeatedly nest bayes rule? And how often must we recalculate p(some event)? If we see a purple giraffe after seeing 100 yellow ones do we change p(purple/giraffe) to ½? Or do we treat this a a rare event and still conclude that most (99%) of giraffes are yellow? How many anomalous events would we have to see to change this? How do we decide how much to change this number by? If it directly effects our well being do we change our probabilities more?

While reading the assigned papers I was reminded of my first attempt to define intelligence. I was taking an independent study course in artificial intelligence, and I had decided to find out what intelligence was before I attempted to create it myself. The definitions I found were very vague. It seemed everyone had their own idea of intelligence. Needless to say, I did not find a concrete definition.

Perhaps my inability to find a definition was the reason that E.G.Boring’s proposal to define intelligence stood out. While I don’t agree with defining intelligence by the measures used in standard IQ tests I do feel that a more fleshed out definition would help focus our efforts to create intelligence. The proposed notion that intelligence is both “the ability to adapt to one’s environment” and “the ability to learn from one’s experience” is one that I generally agree with. However, as revisionists have noted this definition does not address certain factors (such as speed) that we associate with intelligence. Furthermore, there are many ways in which people can adapt or learn. Who is to say that one of those is better than any other?

Sure, defining intelligence, and using that definition to create assessments, would have some advantages. With this information we could identify students who may need help and provide them opportunities that may otherwise not be available to them (remedial classes etc). Unfortunately, anything that can be measured seems to bring with it elitism. If a standard definition for intelligence were to be established today I’m not sure our society would be altruistic. From a employer’s perspective it may seem easier to fire an employee who has a low intelligence score and higher another more “intelligent” employee. From a school administrator’s point of view it may be very tempting to add an intelligence requirement to the admission process. I believe that for many this magic number would be terribly influential. The test would not only measure whatever we defined intelligence to be, but could also determine one’s fate.

Perhaps a single number that embodies intelligence wouldn’t bring about such a dystopia. Still something doesn’t sit well when boiling a person’s cognitive ability down to a single number. For that reason I see myself gravitating towards Gardner’s idea of multiple intelligences. Einstein and Van Gohg are generally seen as geniuses in their own domains. Is there a magic “g” that would have declared them destined for greatness? To me it seems more likely that each had different set of skills that while different were equally impressive.

It is difficult to think of human intelligence as simply one mechanism that that can accurately distinguish about seven inputs. It is also very difficult to think that the only way one can get better is by nesting these mechanisms. I’m not saying that Miller’s magic number has no ground, but I do wonder if he simply discovered a mechanism humans use when they first come in contact with a new problem. When the novices and experts were mentioned I started to think about what makes a computer program more efficient when compared to another. Given a task, one computer program is usually deemed superior if the time for it to execute is faster, if it can keep this speed with a bigger data set, and if it consistently produces accurate results. I would argue that the same criteria can be applied to determine the difference between an expert and a novice. Experts solve problems faster, usually can handle greater data sets, and are rather consistent in terms of their out put (they don’t make many mistakes). In computer programs you generally see a performance boost when a faster algorithm is applied. A substantially modified bubble sort (the worst kind of sorting algorithm save randomly sorting) does not perform much better than a regular bubble sort. Similarly I would like to think that experts develop a structure that is more sophisticated than the 7 input algorithm that Miller proposes we use.

I like the idea of using heuristics to find an optimal solution to a problem. When you have many ways to tackle a problem finding the method that is more useful will definitely increase your performance on a task. In terms of intelligence, this approach also fits with the idea that intelligence is partially defined by how well we can adapt. The problem I see is the generation of these heuristics. Where do they come from? How do we create them? If intelligence is how well we adapt to a new obstacle then surely the generation of these heuristics must be an important component. Even with Miller’s idea of chunking we have to wonder how we create the definition of a chunk. Yes chunking allows us to process more information faster, but at some point biologically, or by some other means, we must identify patterns that we can then use as chunking criteria. For some reason I don’t see the addition of dimensions to be a sufficient answer.

For my story board I chose to represent the task of a player playing king of the hill offensively. There are 3 ways our group discussed in which a player can play offensively. One is of course to score. In order to score players must get to the hill and stay there as long a s possible. The more time on the hill the more points you get. The next way to play offensively is to use virtual weapons. There are 2 types of virtual weapons, bombs and mines. Bombs can be used immediately and affect the surrounding players. Mines can be set at a given location and later detonated from a distance. In all cases players must move to a predefined area whether it be to get on the hill, or to pick up a weapon. These areas are found by searching for their icon in the in game map. Weapons are used/detonated by pressing the weapon’s image on the screen.