What does R^2=1 mean? R^2=0?
R^2=1 means predicted and observed responses are all identical.
R^2=0.64 means that regression explains 64% of the variation, and the remaining 36% is explained by the errors (i.e., the difference between model predictions and observed values).
R^2=0 means variation of observed responses equals variation of predicted responses, so the regression model is no better than simply using the sample mean (rather than an equation) as the model.