Verified by 
Psychology is always called upon to help with the improvement of the person. If psychological science does not rise to the challenge, a lucrative self-help industry is always on hand. Much of what popular self-help books offer, however, is not evidence-based. Many of the remedies are worthless, and some can be harmful. Others that do work may not work for the reasons given. The rise of positive psychology was, in part, a response to the perception that scientific psychology has not met urgent human and social needs. Over 10 years ago, Martin Seligman urged the attendees of a meeting to write science-based self-help books that describe methods that work.
It has been a busy decade for positive psychology and some of its promises have been realized. There is now a lot of research showing that certain positive perceptions, attitudes, and behaviors have beneficial effects on human well-being. Much of this research has been driven by a simple - and for its simplicity appealing - linear main effect model. In correlational research this means that some desirable characteristics (e.g., virtues) are correlated with other desirable characteristics (e.g., health). In experimental research this means that inducing participants to do certain desirable things (e.g. , finding something to be glad about in every situation) will increase desirable psychological states (e.g., happiness).
The linear main effect model assumes that an increase in desirable input will be associated with an increase in desirable output, and that the magnitude of the gain will be the same for any unit of change wherever it may occur on the scale. This is the assumption of linearity. The model also assumes that there are no exceptions to the linear pattern; or if there are, they are few and they are no big deal. This is the main effect assumption. It says that the effect of good leading to good is not qualified by other variables.
But here's the rub. Well, there are two rubs. First, the assumption of linearity is often a mere theoretico-methodological convenience. A correlation coefficient can always be computed. Only a scatterplot and some further testing can reveal if the data are really linear. Although linear effects are indeed common, there are many cases in the psychological literature of there being too much of a good thing. A positive self-image or high self-esteem are nice, but at some point they turn into narcissism. Finding out where that point is turns out to be a difficult task. Confidence also seems like something we all want, until we get into the self-defeating territory of overconfidence. Again, finding where that territory begins is tricky business. Altruism seems like something we all want for society, but now we learn that there is such a thing as "pathological altruism."
Second, a linear relation may hold under certain conditions but collapse or even reverse itself in others. In an American Psychologist article, McNulty & Fincham (2011) argue that positive psychology must acknowledge its boundaries. Some psychological properties or activities conventionally seen as good can have negative effects. McNulty & Fincham provide several examples, but they focus on four findings from their own research on marital satisfaction.
Finding number 1 is that a lack of forgiveness is damaging to satisfaction if there are few hostilities, but not when there are many. In other words, when things are really bad, forgiveness is not much help. This is unfortunate and somewhat ironic because you'd think that forgiveness is most needed in hostile marriages. Finding number 2 is that a lack of optimism about the future is associated with greater satisfaction decrements when the partner has strong rather than weak interpersonal skills. In other words, optimism per se does not do much good. High expectations need to be confirmed, sooner or later, by reality. Finding number 3 is that benevolent attributions (e.g., not blaming the other) are associated with greater satisfaction decrements when marital problems are severe. Again, if reality does not cooperate, it does not do much good to place all one's hopes in a "positive construal." The fourth finding is that a lack of kindness is associated with greater satisfaction decrements when marital problems are mild rather than severe. Again, reality can bite. [1]
In an earlier post, I described a study by Wood et al. (2009), which showed that repetitions of positive self-statements such as "I am a lovable person" increase self-esteem only if self-esteem is already high. If it is low, such statements lower it still. The best intentions cannot guarantee that positive thinking will not backfire.
McNulty & Fincham's move from positive to realistic psychology is an important one. It gets us beyond the simplicity of the linear main effect model, and, more critically, it breaks the hubris of thinking that everything can be fixed if one only puts one's mind into a positive frame. To the extent that positive psychology has allowed this impression to be formed, it has done real damage. McNulty & Fincham do note, however, that several positive psychologists have anticipated these developments, which is to their credit.
Note.
[1] If this paragraph is difficult to read it is because I framed 3 of the 4 findings negatively ("lack of"), which I chose to do because that's where the differences were largest. The attentive reader also notes that there is an overriding negative effect of marriage on satisfaction. Positive attitudes only limit the size of the satisfaction decrements in this research; they do not increase satisfaction. And again, they do so only when things are not all that bad to begin with. See here for an earlier post on the dark side of optimism.
McNulty, J. K., & Fincham, F. D. (2011). Beyond positive psychology? Toward a contextual view of psychological processes and well-being. American Psychologist, 25. doi: 10.1037/a0024572
