Likewise, offering children bribes for good behavior encourages the children to behave well only when adults are watching, while praise without bribes leads to unconditional good behavior.
Parenting tools to shape behavior
Looking at this space of artificial positive rewards to shape behavior, tools include attention, praise, bribes, payments, hugs, gifts, star charts, treats, and more. They all work through similar mechanisms:
Reinforcement propagates backwards through credit assignment. If a child has a tantrum and is offered a reward to stop, and would not have received the reward without the tantrum, then the tantrum is reinforced.
Separately, the strategy of offering rewards to shape behavior is passed on via imitation. The child who is rewarded with gifts is more likely to try rewarding others with gifts. These strategies are then reinforced, or not, based on results.
This is not special to human parenting, we see it in other animals and in non-parenting contexts. It doesn’t give us much reason to expect bribes to result in deception and praise to result in “unconditional good behavior”.
Value of parenting research
While there is research comparing specific parenting tools, it’s weak evidence:
Observational studies have lots of confounding factors.
Children vary in how they respond to parenting tools.
Parents vary in their comfort and skill with parenting tools.
It’s hard to measure how people act when they are not measured.
RCTs are limited, we can’t force parents to use specific tools more or less often.
There are biases in the research process and many results don’t replicate.
Such evidence as I’ve seen filtered through books and other sources suggests that the type of praise can help direct the credit assignment process. So we get the advice to praise the specific behavior we want to reinforce.
I wouldn’t be surprised (20%) if this fails to replicate, but I give it some influence. If so, praise is a weaker but more targetable reinforcer, whereas (eg) candy is a stronger but less targeted reinforcer.
Does Santa bring bribes or praise?
Gifts, obviously. But also praise. Santa brings gifts to “good children”, so if a kid gets a gift from Santa then it follows they are a “good child”, and that is praise. And if a child eats their vegetables on Christmas Eve and their parents say “Santa will be happy you’re eating healthy” then that’s specific praise.
What about The Third Alternative?
As hinted above, parents use a lot of tools to raise their children. There are multiple forms of artificial positive rewards and multiple ways to use them, and that is just one general category among many. The tools aren’t mutually exclusive. There is no shortage of alternatives, including wild and creative brainstormed options.
So instead of choosing between Alternative One and Alternative Two in a False Dilemma, the actual problem is choosing a portfolio of tools based on their situation, trying to find the Pareto frontier based on their goals and values.
Disclaimer: I am not arguing for or against “Santa-ism”. Policy Debates Should Not Appear One-Sided. I am instead interested in parenting in general.
Parenting tools to shape behavior
Looking at this space of artificial positive rewards to shape behavior, tools include attention, praise, bribes, payments, hugs, gifts, star charts, treats, and more. They all work through similar mechanisms:
They reinforce behavior; “reward chisels cognitive grooves into an agent” (from Reward is not the optimization target).
Reinforcement propagates backwards through credit assignment. If a child has a tantrum and is offered a reward to stop, and would not have received the reward without the tantrum, then the tantrum is reinforced.
Separately, the strategy of offering rewards to shape behavior is passed on via imitation. The child who is rewarded with gifts is more likely to try rewarding others with gifts. These strategies are then reinforced, or not, based on results.
This is not special to human parenting, we see it in other animals and in non-parenting contexts. It doesn’t give us much reason to expect bribes to result in deception and praise to result in “unconditional good behavior”.
Value of parenting research
While there is research comparing specific parenting tools, it’s weak evidence:
Observational studies have lots of confounding factors.
Children vary in how they respond to parenting tools.
Parents vary in their comfort and skill with parenting tools.
It’s hard to measure how people act when they are not measured.
RCTs are limited, we can’t force parents to use specific tools more or less often.
There are biases in the research process and many results don’t replicate.
Such evidence as I’ve seen filtered through books and other sources suggests that the type of praise can help direct the credit assignment process. So we get the advice to praise the specific behavior we want to reinforce.
I wouldn’t be surprised (20%) if this fails to replicate, but I give it some influence. If so, praise is a weaker but more targetable reinforcer, whereas (eg) candy is a stronger but less targeted reinforcer.
Does Santa bring bribes or praise?
Gifts, obviously. But also praise. Santa brings gifts to “good children”, so if a kid gets a gift from Santa then it follows they are a “good child”, and that is praise. And if a child eats their vegetables on Christmas Eve and their parents say “Santa will be happy you’re eating healthy” then that’s specific praise.
What about The Third Alternative?
As hinted above, parents use a lot of tools to raise their children. There are multiple forms of artificial positive rewards and multiple ways to use them, and that is just one general category among many. The tools aren’t mutually exclusive. There is no shortage of alternatives, including wild and creative brainstormed options.
So instead of choosing between Alternative One and Alternative Two in a False Dilemma, the actual problem is choosing a portfolio of tools based on their situation, trying to find the Pareto frontier based on their goals and values.