Based on my reading of the book, I guess the main suggestions that I’d make to anyone interested in playing around with creating new self-concepts would be:
Use lots of diverse examples. E.g. if you’re creating a self-concept of kindness, look for memories of both large and small acts of kindness. The memories contained within your self-concept serve as kind of templates to match various experiences against; the more diverse the database of templates, the more likely it is that different actions will be correctly identified. (Machine learning folks might say that this avoids overfitting.)
You can try to explicitly include memories from different years of life, as well as covering both short and long time periods (e.g. if you hold the door open to someone, that’s an act of kindness lasting for a few seconds; if you have a friend who you’ve been helping with their troubles for the last several years, that can be thought of either as a big set of short acts or a single long-lasting act).
See if you can include counterexamples as well, maybe doing a negative-positive reframe to turn them into examples, or including qualifiers with them. Again the template matching thing: if your self-concept for a quality contains instances of occasions when you failed to act according to the quality, then that allows you to recognize future occasions when you’re not acting in accordance to the quality. This helps avoid unrealistic overconfidence in the quality.
(And for anyone interested in doing this seriously, I much recommend the book for lots and lots of practical examples and tips.)
Based on my reading of the book, I guess the main suggestions that I’d make to anyone interested in playing around with creating new self-concepts would be:
Use lots of diverse examples. E.g. if you’re creating a self-concept of kindness, look for memories of both large and small acts of kindness. The memories contained within your self-concept serve as kind of templates to match various experiences against; the more diverse the database of templates, the more likely it is that different actions will be correctly identified. (Machine learning folks might say that this avoids overfitting.)
You can try to explicitly include memories from different years of life, as well as covering both short and long time periods (e.g. if you hold the door open to someone, that’s an act of kindness lasting for a few seconds; if you have a friend who you’ve been helping with their troubles for the last several years, that can be thought of either as a big set of short acts or a single long-lasting act).
See if you can include counterexamples as well, maybe doing a negative-positive reframe to turn them into examples, or including qualifiers with them. Again the template matching thing: if your self-concept for a quality contains instances of occasions when you failed to act according to the quality, then that allows you to recognize future occasions when you’re not acting in accordance to the quality. This helps avoid unrealistic overconfidence in the quality.
(And for anyone interested in doing this seriously, I much recommend the book for lots and lots of practical examples and tips.)