Effect or Predictor(s), is it always clear which is which? How does one determine if a variable is an effect or a predictor if all the variables are measured and none or manipulated or otherwise fixed? For example, does reduction in vascular diameter cause increased amyloid accumulation or does amyloid accumulation drive changes in vascular diameter?
There's no way to explicitly model this question if what one has is just data from brains, as far as I know, but I don't know much.
 A: There are several questions here to be disentangled. They sometimes have the same answer, but far from always.
What do you want to predict? Sometimes the interest is in predicting an outcome from predictor variables, as when the interest is in predicting the growth rate of an economy, the level or discharge of a river, or children's heights as they grow. But often the interest is reversed. If a skeleton is found, our idea is that skeletons grow as people do, and so that age and getting older are causal, but in practice we may use bone length to predict age, which is unknown but of great interest.
What is considered a response or outcome? Here language may vary, but very often there are detailed ideas (in science, engineering, medicine, social science, and so forth) about how something varies in terms of associated causal mechanisms, processes, behaviour, or whatever analysis appeals to people in a field. Thus wheat yield might be a response to temperature, rainfall, soil nutrients, labour and materials inputs, and so on, but most of these would not usually be considered to be affected by wheat yield. (On the other hand, a growing crop can deplete nutrients.) Sometimes time is of central concern in that causes supposedly precede effects in time and that is indeed often how causes and effects are distinguished. But people and organisms can engage in anticipatory behaviour, as when presents are bought or made before the time of giving, or animals prepare for seasonal changes.
Is there a experimental set-up? Typically controlled variables are regarded as causal, say temperature or pressure in chemistry or engineering, and consequences, say output of products, as the outcome or response variable.
Are variables on the same footing? Variables may be loosely or even strictly regarded as having the same status. Height and weight might be considered thus, or students' performance on several different examinations. Nothing is absolute here: examination performance might be analysed in terms of measures of ability, student effort, personal or family background, and so forth, but the examinations themselves might not be independent, as when one examination follows another quickly, and many students don't revise enough for the second.
Are there feedback loops? The presence of a feedback loop can throw doubt on which of two variables is cause and which effect and one good answer may be neither.  The analysis may then depend on the time scale of interest. At any moment a land surface may be considered fixed and to determine or at least influence the rate at which it changes, as when the movement of sediment is predicted from slope or other topographic measures. But wait long enough (geological time) and the land surface itself changes as a result of those sediment movements.
