Bryan Caplan  

Praise: Substitution versus Income Effects

Hopeless Excuses... On Meddling...
Touchy-feely parents shower praise on their kids.  "Great job!"  "You're super smart!"  "Wonderful."  Old-school parents do the opposite.  "You could have done better."  "A-?"  "That won't get you into Harvard."

Why the chasm?  The real story, I suspect, is emotional rather than strategic.  Parents praise or withhold because that's what feels right to them.  The charitable story, though, is that strategy is central.  The two archetypes factually disagree about the effect of praise on performance, and act accordingly.

The pro-praise story: Praise is a form of reward.  The greater the rewards of success, the more effort kids exert.

The anti-praise story: Yes, praise is a form of reward.  But the more rewards kids rack up, the more satisfied they feel.  The more satisfied they feel, the less effort kids exert.

Framed this way, the pro- and anti-praise debate boils down to the intermediate micro analysis of the substitution and income effects.  Does paying people more (or taxing them less) make them work more or less?  The strange but true answer is: It depends.  "The greater the rewards, the greater the effort" makes sense.  But so does, "The more rewards you have, the less you crave further rewards."  The great "to praise or not to praise" debate fits elegantly into this framework.

Or does it?  Consider: Touchy-feel parents also typically avoid shaming their kids.  Old-school parents, in contrast, shame freely.  Here, then, old-school parents seem to rely on the substitution effect - the greater the cost of bad behavior, the smaller the quantity.  Touchy-feely parents, in contrast, seem to tacitly appeal to the income effect: A shamed kid will act even worse because he has so little left to lose.

Personally, my parenting style embraces the substitution effect in both directions.  I happily praise good behavior, and sternly (though not angrily) criticize bad behavior.  That's definitely more consistent than either of the classic archetypes, and seems to work well for my kids.  But perhaps that's an illusion - or an outlier.  From a bird's eye view, which view of praise and blame has the facts on its side?

COMMENTS (18 to date)
An onyx mousse writes:

[Comment removed pending confirmation of email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Chris Thomas writes:

The above commenter is correct that the behaviorism literature has a lot to say on this. In general, reinforcement is more effective than punishment, and has fewer negative side effects. There is an emphasis on using a variety of reinforcers to avoid satiation. And as for ignoring bad behavior, behaviorism always takes a function-based approach, i.e., you choose how to address behavior based on why it is happening. Bad behavior that is aimed at escaping demands will likely be strengthened when ignored, bad behavior aimed at getting attention will likely be weakened when ignored, and bad behavior done for its intrinsic pleasurableness will likely be unaffected when ignored. A typical approach would involve providing reinforcement for existing good behaviors, and teaching and reinforcing appropriate alternatives to bad behaviors. All this while regularly making sure you are using varied and effective reinforcers, e.g., verbal praise is unlikely to be an effective reinforcer for all behaviors, all of the time. For further reading, Behavior Anslysis for Lasting Change and Applied Behavior Anslysis are the standard text books, and the Journal of Applied Behavior Analysis is the main academic journal.

Xerxes writes:

If only I had any faith at all that the findings from behaviorism reflected reality.

Matt Moore writes:

This is one of the classic examples in Thinking Fast and Slow. Criticism seems more effective than praise due to reversion to the mean. To the extent that we suffer from this attribution error, we should have a lower threshold for providing praise.

However, that's tangential to the discussion. I think that these archetypes are maximising for different outcomes. Giving a lot of praise will tend to yield confidence, whereas criticism will yield motivation. Both are needed to some degree, with the underlying character of the child being important in determining the parenting strategy.

A final wrinkle. 'Studies show' (no reference to hand, sorry) that the type of praise matters very much. If you praise the child for being smart, or another trait, you might indeed make them lazy and challenge averse. However, if you praise them for effort, or other behavioural aspects, then you do indeed increase motivation.

Michael Savage writes:

Another way of framing the difference is around duty. Traditional parents tell you to get good grades, and go call on your grandmother because it's your duty. That leaves a space of psychic freedom - you can rebel, not want to do it, but still conform outwardly. Touchy-feely parents want you to *want* to get good grades so you can have better life choices, and call on your grandma because she'll be so sad if you don't see her. That gives another perspective on allocation of shame. Both allocate shame. But being ashamed of not doing your duty is maybe more easily absorbed that being ashamed of who you are, ashamed that you don't want the right things. (this from Zizek, who makes some interesting points amid the impenetrable craziness)

Tiago writes:

I would add that criticism should be significantly better than praise for people to adopt it, since praise creates happiness and criticism unhappiness.

Dan writes:

In my experience in raising genetically diverse children is that it depends greatly on the child. Children have varied dispositions and as a result respond very differently to criticism, praise, discipline, and reward. They also often behave very differently at school than at home. Generally, "studies" and other research is of limited value in specific individualized situations, as the studies aggregate.

Steve Horwitz writes:

I would largely agree with Bryan's approach here. My only concern about blame/criticism is that I think it's counterproductive when it's for the failure to reach some externally imposed standard rather than the parents' own expectations of their children.

I would have never criticized my kids for not getting an A because it wouldn't get them into Harvard. If I thought they had been lazy or the like, then criticize their performance, because I expect them to do their best. But I never expected them to get into Harvard (or wherever).

Parenting is full of local knowledge. You know your kids best and you know what they're capable of and good parents communicate that they understand their kids and that they have certain clear expectations. Freely criticize them for not doing what you expect and what they are capable of. Don't criticize them for falling short of perfection or someone else's notion of ideal.

And give where praise is due. Happily and freely.

guillermo writes:

The Effects of Praise on Children’s Intrinsic Motivation:
A Review and Synthesis

Brian Holtz writes:

Praise liberally but honestly. Shame sparingly. Criticize constructively, never angrily. In lieu of using anger as a punishment, I levy "butler points" as fines on misbehavior. I redeem each butler point for 1-2 minutes of effort at my convenience e.g. fetching me something from the other end of the house. Butler points are tracked as pictures on my smart phone, and are deleted on redemption.

An ensuing controversy is whether I should be allowed to levy a butler point in lieu of (or in addition to) inter-sibling reparations for inter-sibling infractions. I say that reparations are insufficient, because I seek to minimize the aggregate incidence of misbehavior, and not merely the imbalance in it.

RPLong writes:

Here's the Last Psychiatrist's take (link here - warning: spicy language):

Slightly off topic but here's an important example: say you yell every day at an/your eight year old girl for sloppy homework, admittedly a terrible thing to do but not uncommon, and eventually she thinks, "I'm terrible at everything" and gives up, so the standard interpretation of this is that she has lost self-confidence, she's been demoralized, and case by case you may be right, but there's another possibility which you should consider: she chooses to focus on "I'm terrible at everything" so that she can give up. "If I agree to hate myself I only need a 60? I'll be done in 10 minutes. "
It is precisely at this instant that a parent fails or succeeds, i.e. fails: do they teach the kid to prefer (find reinforcement in) the drudgery of boring, difficult work with little daily evidence of improvement, or do they teach the kid to prefer (find reinforcement in) about 20 minutes of sobbing hysterically and then off to Facebook and a sandwich? Each human being is only able to learn to prefer one of those at a time. Which one does the parent incentivize?

In the end, I think we should lavish children with praise when they accomplish something new, and then move on. I'm very proud of you for learning your ABCs, Timmy. Can you tell me how to spell dog? Everything is a fun accomplishment, and there is always something new around the corner. It's more important, in my view, to reinforce the fun of learning and the happiness of accomplishment than it is to use rewards or punishments to extract particular outcomes.

Mark S writes:

Sal Khan of Khan Academy never tells his son he's smart:
I decided to praise my son not when he succeeded at things he was already good at, but when he persevered with things that he found difficult. I stressed to him that by struggling, your brain grows. Between the deep body of research on the field of learning mindsets and this personal experience with my son, I am more convinced than ever that mindsets toward learning could matter more than anything else we teach.

David Condon writes:

Both of your examples are of poor parenting decisions. Parent type 2 are using coercion, which is generally only useful in the short-term. Parent type 1 are using incorrect corrective feedback. Phrases like "you're very smart" are known variously, depending on the field, as:

Education - praise
Social Psychology - fixed-mindset
Behavior Analysis - non-contingent reinforcement

More specifically:

Praise has an inconsistent definition in the field of education (which is why I think it's stupid that anybody ever uses it as a technical term, but whatever), but is frequently referred to as feedback which makes reference to the self.

This definition of praise is consistent with social psychology's definition of fixed-mindset as feedback that makes reference to previously acquired attributes as opposed to referencing growth of new attributes.

Non-contingent reinforcement is a broader term referring to a process of reinforcement without making any reference to the targeted behavior.

All three disciplines are more or less in agreement here. If the goal is to teach them how to spell, telling them they're smart rather than they did a great job spelling that word is counterproductive.

I think the best paper on this issue is Kluger & DeNisi 1996 The effects of feedback interventions on performance

However, I'm not such a fan of some of their theoretical explanation of their results. I agree with Chris Thomas that behavior analysis is the best source for understanding the underlying theory. Specifically, understanding how to use feedback effectively requires a combined understanding of Differential Reinforcement, Noncontingent Reinforcement, Schedules of Reinforcement, functional analysis, and the Matching Law, and how they relate to the problem.

The use of punishment, or a process which decreases problematic behavior, is probably underutilized, but the notion here is that too much punishment is a bigger issue than too little punishment, so it's generally avoided in favor of extinction. This is probably a mistake in many cases.

Xerxes, I believe you're confusing disagreements over behaviorist philosophy with disagreements over behaviorist methods. Although arguments over the former are quite common, behaviorist methods are quite possibly among the least disputed in all of psychology. Very few researchers argue that operant or respondent conditioning doesn't work at this point.

ChrisA writes:

There are obvious parallels here between children and employees. It would seem quite valuable if we could figure out how to best motivate workers to do their best. But I don't see a consistent approach between firms or even within companies. Some bosses are highly critical and sparse in their praise, famously Steve Jobs. Others are very generous in this area. My view is that people compensate for the authority figure calibration and make internal reference to that. So a small hint of praise from Steve Jobs means the same as a lengthy speech of praise from a typically more generous boss. Same with kids. If you usually only criticize then a scant word of praise means a lot in terms of providing needed reinforcement of desired behavior and vice versa. I would suggest inconsistent approach is the worst method of motivation, since that is most likely to confuse.

Nathan W writes:

I'm not sure why you associated "touchy feely" with positive reinforcement as a motivational strategy for study and learning. What you're talking about is positive reinforcement, unless you're intending to refer to parents who offer praise regardless of whether the underlying achievement or effort is of low or high quality.

With respect to the question "which view of praise and blame has the facts on its side?", I take a similar view to related questions regarding teaching style, etc.

Children can be pretty good at spotting fakes sometimes. I don't think they'd get much out of fake praise from a parent who wasn't "naturally" inclined to this approach, but who generally adopts it as a strategy because they have been told that it works better. And more specifically, the parent will be best, and most consistent (and I argue that consistency is probably more important than the strategy itself), in applying the strategy which feels most natural to them.

Similarly, I think teachers should be broadly free to adopt their preferred teaching styles and methods, because they will be comparatively ineffective in applying a strategy that does not suit their personal character very well.

Three other notes:
1) I think the strategy of using both positive and negative reinforcement (Chris's caveats about bad behaviour seem very good here too) will be the best strategy in the vast majority of cases.
2) Every child is different. Over the years, a parent experimenting with diverse strategies for motivation and discipline will find what works best. They should be open to the whole range of strategies in determining what works best according to their own character and what is most effective in leading the child towards the desired result, which I assume can basically summarized as high intrinsic motivation and good self discipline.
3) The connection between the behaviour/outcome and the positive/negative reinforcement needs to be exceedingly clear. With young children, this means the positive/negative reinforcement needs to be very closely connected in time. For example, if you come home and the babysitter tells you that a 5 year old behaved badly several hours ago, the punishment is basically just mean. But for an older child, when providing the positive/negative reinforcement, the reason for it should be clear in each instance.

Finally, I agree very strongly with ChrisA about inconsistency. Even a little bit of inconsistency can ruin everything.

Jerome Bigge writes:

Children who fell they can never meet the standards set for them will eventually give up trying to do. Their attitude is "simple": I can never be good enough, so why bother?

Ted writes:

Motivating children, as with employees, is best achieved with a balanced approach of praise and criticism calibrated for the individual. Shame is a necessary factor, but must be precisely calibrated to the individual or undesirable results ensue.

Where I take issue with the "touchy-feely" parenting paradigm is the laziness involved in embracing faddishness. Rather than attentive calibration informed by thoughtful analysis, the intellectually bankrupt parent seeks and utilizes the "blueprint du jour" as a means of abdicating their responsibility to correctly evaluate the needs of their children.

I have endured the results of such abdication, in the workplace, for the last thirty years. Too often, the product of motivational faddishness is entirely unprepared for the hard realities of the adult physical world. It is (arguably) more difficult for those entering the workforce in physical-reality-based occupations where the results of failure are universally discernible, whereas the arena of so-called "knowledge work" appears to provide, in large measure, an effective milieu for the perpetuation of motivationally crippled individuals.

Glen Smith writes:

Most of the time I've been praised by an employer, I was most definitely not doing my best while most of the time they were dissatisfied with my performance, I worked as hard as I could.

Comments for this entry have been closed
Return to top