Aristotle does not limit motion to change of place, to growth and decay, to alteration, or the like; for motion, while it encompasses these things, cannot be thought of as one sort of motion that all other sorts of motion can be reduced to (i.e., motion cannot be alteration, while all other forms of motion can be reduced to alteration). Neither can the sorts of motion, added together, tell us what motion itself itself is -- any more than listing different virtues can answer the question of what virtue itself is -- and so Aristotle must give an account of motion that goes beyond listing different sorts of motion, or collapsing different sorts of motion into a single kind of motion; or to put it another way, Aristotle must explain motion as such.
In III:1 of the Physics, Aristotle defines motion as "the being-at-work-staying-itself of whatever is potentially, just as such" (201a10-20) and again as the "being-at-work-staying-itself of what is potentially, whenever, being fully at work, it is at work not as itself but just as movable" (201a15-30). The definition might be formulated in a more paradoxical (and troubling) way: motion is the activity of potentiality; and thus one might conclude Aristotle's definition directly contradicts itself, for actuality and potentiality ought to be opposed to one another--at least on the superficial reading.
How would a potentiality be at work while remaining as a potentiality? Isn't potentiality precisely that which has not yet been actualized? Aristotle clarifies his definition by saying that he does not mean that a particular potentiality for a particular being constitutes motion, else we might say that to be brown constitutes motion; rather, motion is the activity of potentiality as potentiality. To take an example, a light-skinned person who rarely spends time in the sun can be called potentially tan, and while "being tanned" obviously does not constitute motion in its essence, neither -- precisely speaking -- does the becoming tan from being light-skinned constitute motion itself (though it is a motion). What, in the process of becoming tanned, constitutes motion itself? In any particular motion, motion itself must be present, accessible to us on reflection, and since we know that motion is the activity of potency as potency, we can ask: how, when a light-skinned person becomes tanned, does motion manifest itself as an activity of potency?
Motion cannot be inextricably tied to any particular potentiality, but it must be related to potentiality as such; and so we say that motion manifests itself in the fact that, while one becomes tanned, being light-skinned slides from actuality into potentiality. By always claiming, as it were, one contrary, and giving the other contrary to actuality, potentiality always keeps something for itself; for contraries cannot both be actual at one time (a person cannot at the same time be both light-skinned and tanned). Motion is the activity of potentiality's maintenance of its own reserve, and this can only be possible with finite beings. Motion does not consist in being light skinned or in being tanned, but in necessarily only being one at the same time, while potentially the other; and one cannot help but note that even potentiality, considered in itself, exists as an actuality.
This brings us back to the previous insight that potentiality and actuality cannot be opposed to one another, that potentiality, in order to be at all, must be actual; yet, in spite of this, one cannot simply conflate potentiality with actuality--the two must be considered distinct, although connected. In order to see the distinction, we must return to the problematic definition that seems to threaten the distinction between actuality and potentiality: potentiality, to be potentiality, must be actualized in the structure of motion. This can be reformulated as: motion is potentiality being itself actually. Aristotle goes a step further, defining motion in explicitly contradictory terms as an "incomplete being complete" (257b8-10), but rather than creating an impasse, when Aristotle defines motion in its most problematic form, he opens the way to altering the nature of the problem. Motion itself may be called a complete way of being incomplete, an actuality that preserves potentiality, not because motion cannot ever be abstracted from things (though it must always exist in them), but rather because it constitutes the being of composite things. The definition of motion as an incomplete being complete forms a bridge, for it points in two directions: motion in itself as the active maintenance of a reserve of potentiality (the potential maintained being the incomplete), and motion as always making possible an actual, composite being. The two can only be separated in thought; for to be a composite being means to never be fully complete (i.e., fully actual), and to be a motion means to be completed only in the presence of a composite being that takes responsibility for the motion (as the first mover).
Motion's completeness is the composite being's incompleteness; this only says that motion finds its explanation in a being that takes responsibility for it, and the composite being only finds its being composite in the perpetual presence of the potentiality that motion preserves. Here a problem arises, for it seems as though I might be using "motion" equivocally: first as a particular motion, and second as motion itself (the structure of motion as motion). A particular motion, in order to be understood, requires a particular being to take responsibility for it (to be its first cause), but it does not seem as though composite being as such owes its being-composite to a particular motion (becoming tanned, changing location, and so on), but rather to motion as the preservation of potentiality. Note, however, that although motion finds its explanation only in a being that takes responsibility for motion, this does not immediately indicate that this being must be composite or that the motion is a particular motion. One can read Aristotle's definition of motion as creating a fissure in the ontic through which the ontological makes its appearance: particular motions may be explained in one way by reference to natural beings responsible for motion, while they may be explained in another way by motion as such; however, motion as such cannot be explained by particular motions, nor by reference to itself, but must be explained by its causes. Thus, the question of the cause of motion as such lies implicit in both particular motions and the structure of motion that makes composite beings possible. An important question arises at this juncture: does motion (not particular motions, but motion as such), which gives composite beings their being as composite, belong properly to the composite being itself, as an aspect of that being, or does it belong to something higher than the composite being itself, being given from another source (and it seems clear this source would be the cause of motion)? To put it a clearer way: can we call motion as motion prior to the composite beings in which motion constitutes their being as composite, or are composite beings ontologically prior to motion as motion? Can we call composite beings prior to their being, or must we call the being of composite beings ontologically prior? This line of questioning interrogates the possibility of the presence of the transcendent in the immanent as well as the basic character of their ontological relation. How Aristotle answers the question determines how he conceives of the ontological difference.
Aristotle begins his explanation of motion as such by first considering particular motions; considering the ontic first, to reach the ontological. The cause of motions must be explained by reference to a first mover that exists as a particular being, and so we would suspect that the question of the cause of motion itself might move along similar lines; as said above, motion cannot be explained by particular motions, and so its cause must lie in itself, or in an external cause. If the first cause of motion itself (first, of course, in the ontological, not temporal sense) turns out to be a composite being, then composite beings would be ontologically prior to motion itself, and this would be the case if motion exists simply as a feature of composite beings.
In order to determine whether the cause of motion as such can be understood to be ontologically determined by motion itself, Aristotle moves back to the consideration of particular motions so that he might determine whether individual motions encompass their causes (that is, whether the causes of motions are themselves moved by the motion they cause). Some things seem to be obviously moved by something else, such as when a man moves a rock with a stick. In this case, the first mover must be said to be the man, not the stick, for the man bears responsibility for the motion of the rock, and the stick serves merely as an instrument; further, the motion of the rock, though caused by the man, did not move the man himself (the motion of the man moving the rock must be distinguished from the motion of the rock itself). From this, we might be led to conclude that the mover causes the motion, but remains outside of that motion. However, in some cases, it seems as though the moved thing moves itself. The man who moved the rock perhaps looks about for a stick to move the rock with, and this motion seems to originate within the moved thing itself (in this case, the man). Aristotle regards animals as paradigmatic cases for beings, and so the seeming self-motion of animals poses a particular problem if he wishes to maintain that the mover causes the motion without being moved by it. In order to establish that the mover must be unmoved by the motion caused, Aristotle must show that a self-moved mover in some way must be unmoved by the motion.
A self-moved mover must either move itself as a whole, or some part of the self-moved mover must move the whole mover. However, if a whole moves a whole, then the distinction between mover and moved collapses, for that which bears responsibility for the motion also undergoes the motion, and the causing motion and being caused are not two separate things, but one in the same; and -- as Aristotle points out earlier -- if it is possible to collapse the distinction between mover and moved, then teaching and learning could be the same--this cannot be true. An even more basic (though similar) problem arises if one says a whole moves itself as a whole: in order for a whole to be moved, it must be first potentially movable, then brought into motion by an actuality; but if the whole is moves itself, then it must be both be both potentially and actually in the same way at the same time. Therefore, a whole cannot move itself as a whole.
It follows that in a self moved thing some part must move the whole, but that part can move the whole in two ways: either by being moved itself, or by remaining unmoved; but if the part moves itself with the motion it causes in the whole, then the part may itself be viewed as a self-moved mover, in which mover and moved must again be distinct. Thus, the self-moved part that moves the whole can itself be viewed as a whole that must also be divided into a part that moves and a part that is moved, and so the only way to avoid an infinite regress would be to identify a part that moves the whole, but is itself unmoved. Since the part that causes motion does so not in a temporal sense, but in an ontological sense wherein that part causes motion by being primarily responsible for it, it would be irrational to say that there is nothing responsible for a motion. Therefore, even in self-moved movers, an unmoved mover causes the motion, and so the cause of motion in all moved things itself remains beyond the motion caused. Taking this logic from the ontic to the ontological, it follows that the cause of motion as such cannot itself be determined by motion, but must be outside composite being. Therefore, the being of composite beings, motion, is ontologically dependent not upon composite beings themselves, but upon a being that is beyond potentiality, and is itself purely actual.