potato comments on The Power of Reinforcement - Less Wrong

96 Post author: lukeprog 21 June 2012 01:42PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (467)

You are viewing a single comment's thread.

Comment author: potato 21 June 2012 05:55:31PM 1 point [-]

Does this still work if I reinforce myself? Every time I read 5 lesswrong articles in a day, I give myself a reward. Or every time i have a cigarette, I kick a brick wall with no shoes on. If i was consistent with this for a long time, would it work?

Comment author: wedrifid 21 June 2012 06:25:28PM 7 points [-]

Or every time i have a cigarette, I kick a brick wall with no shoes on. If i was consistent with this for a long time, would it work?

Totally. The wall will fall over in 20 years, tops!

The actual answer is maybe - it works for some but not others. A common point of failure is that people just train themselves to cheat and take the reward anyway. I'm not sure what the response rate is when full compliance to the reward schedule is assumed.

Comment author: TheOtherDave 21 June 2012 07:04:33PM 1 point [-]

It can. Basically the failure modes are the same as when reinforcing others. In particular, it's common to fail to maintain consistent thresholds of self-reward.