Heather Vescent

Failing Pattern Recognition

August 27, 2011 · Forecasting, Future, Introspection

I used to wonder why my dog seemed to know when the boyfriend (aka S.) got home while we on a walk. At a certain point in the walk, he would stop being interested with the walk, sniffing the bushes or wanting to walk more and instead - huff it back home. Mr Dog would practically drag me the last block home, breaking into a run if possible, down the driveway and up to the front door. 

Yeah, sure, sometimes S. would be home; but sometimes he wasn't. Mr Dog was never upset when he ran back and there was no S. When Mr Dog ran back and S. was home = positive reinforcement that maybe Mr Dog has some secret sense. But he wasn't always right. And when S. wasn't usually home when Mr D and I  got home, Mr Dog stopped dragging me that last block.

I never really got it. Until last week.

I read some research or an article (I forget exactly where - probably here on the internet or in one of my Science/Nature/Science Mags subscriptions) that explained, when a rat was randomly rewarded with a treat 60% of the time he pushed on a bar. He would push it all the time, even the times he didn't get a treat. This helped him maximize the treat - sure he only got it 60% of the time, but that 60% of the time was worth the 40% he didn't get one. 

The study compared this experiment to a similar one done on humans. In the human case, the human tried to figure out what the pattern of reward was to increase the reward beyond 60%. In doing so, the human actually reduced the successful reward to 40 or 50% (I don't remember off the top of my head). So the human, in trying to figure out the pattern, actually reduced it's success.

That seemed to explain my dog's behavior.

Then I started thinking about my behavior - how I tried to understand and recogonize patterns. What if all those so called patterns are actually pure randomness? It's changed the way I've been thinking about some things. 

I think I'm onto something here. I think getting caught up in patterns: trying to identify them, trying to understand them, might not always be so helpful. (Now, of course, there are times/things that do have patterns.) Maybe not everything has a pattern. And maybe some of the more complicated patterns we believe in, may be random or wild cards in retrospect.