You should have a formula in your book for calculating confidence intervals. From what I remember these are the elements you need to calculate a confidence interval:
1) a value (like the mean)
2) the standard error of the measure
3) the desired width of the CI
You can decompose the 90% CI that you were given to find some of those elements (the CI, the standard error) to get to the mean. With this new mean you can then recalculate your 95% CI with .08 standard error.
Does that make sense?