Is feedback a verbal reward or a tangible reward?

Summary

Decci, Koestner and Ryan (1999) examine the effect of various kinds of rewards on behaviors that people already have some intrinsic interest in doing. They divide rewards according to the following taxonomy: 

Verbal

Tangible

    - Unexpected

    - Expected

       - Task non-contingent (reward is given no matter what)

       - Engagement contingent (reward is given if people do the task)

       - Completion contingent (reward is given if people finish the task)

       - Performance contingent (reward is given if people meet certain performance expectations)
 

 Deci et al. find that verbal rewards have a positive effect on people's motivation (providing that it is seen as informational rather than controlling) but that tangible rewards have a negative effect (unless unexpected or non-contingent). Finally they note that performance-contingent rewards have a particularly negative effect if people do not receive the maximum possible reward.

There are two explanations for this effect. Deci's CET explains these findings by noting that rewards are often double-edged swords. On one hand, being rewarded feels good because it assures people that they are living up to the standards of a community. On the other hand, being rewarded feels bad, because it undermines people's own feelings of autonomy. The alternative explanation is an attributional account. Essentially, in reward conditions people misattribute their intrinsic enjoyment of a task as caused by the reward, leading them to conclude that they are in fact not interested in the task at hand. At the time of publication, the data did not allow these two accounts to be pitted against each other.

Criticism

There are a number of ambiguities about why verbal rewards function differently than tangible rewards (e.g. the degree of expectedness, informational content etc.). The supplementary analysis on that looked at informational versus controlling verbal rewards was a useful step towards addressing this question, but it seems to me that many of the categories used to describe tangible rewards (non-contingent --> performance contingent) could also be used to classify verbal rewards. This would allow a more careful analysis of the interaction between reward type and contingency. Deci et al. only had 21 verbal reward studies to work with, so this was probably not possible at the time, but someone should follow up on this.

Application

This paper left me wondering whether feedback is best considered a verbal reward or a tangible reward, or what their model would have to say about feedback. Intuitively, feedback feels more like a verbal reward, in that it is the direct communication of performance feedback rather that an end in itself. However, it does have a certain physicality, permanence and economic implication inherent in it as well (see Friedman).