Readers, beware! This is a dense bit coming up. However, it is one of the greatest of all passages on this issue written by a Thomist.
"Now, if we conceive of the image [i.e. phantasm] acting on the understanding as moved by the divine mind as principal cause, can we attribute to the image any efficient causal power of its own? [Yes!] This condition is absolutely essential to the notion of the instrument, and if it is lacking, then properly speaking, the role of the image in the production of the idea is not really an instrumental role. We call the image an instrument, writes an author who has sounded these problems in their depths, because in dealing with these mysteries, we speak as best as we can. [Cf. Cajetan, In Ia q.85 a.1; John of St. Thomas, Phil. Nat. IV, q.10 a.2 (Reiser III 306A11); St. Thomas, Quodlibet VIII a.3.]"
"Accordingly, we must conclude that although the instrumental function very closely resembles the function performed by the image with respect to the mind, they are not exactly the same. Trying to be more positive and precise, we may, first of all, observe that although an instrumental function has to do more with the order of efficient causality, the actual function of the image is more in the order of specification and objective causality. An instrument, as happens in the examples we have given, may play a specifying role. But such a role is not of the essence of the instrumental function; in the supremely typical—and supremely pure—case of the instrument taken in hand by the divine Omnipotence, this specifying role may disappear completely. Thus, rather than its possible specifying role, what defines an instrument properly so-called is an effective quality of its own for which it is used by the principal agent. The case of the image, however, is different. Its primary function is to contribute to the specification of the intellect, to put the intellect in touch with its object, and if in doing this the image has to exercise a certain efficiency on the mind, the sole reason for it is to specify the minds own action."
"This primacy of the specifying role goes hand in hand with a reduction of its efficient role, since the image has, with regard to the intellect, no active power distinct from what the mind confers on it. And so, while for an instrument properly so-called efficiency comes first and specification comes second, for the image it is the other way around. In matters of knowledge, the main thing is not to produce an effect but to exist with a particular specification; efficient causality is regularly subordinated here [i.e. in intellection] to specification, which it serves as a condition and means. A more precise understanding, then, of the causal function of the image in the production of the idea could perhaps be worked out with the notion of a formal or objective instrument, in which the principle of instrumentality would be transferred from the order of efficient causality to the order of formal causality and in which efficiency would remain only as a connotation."
"[In a footnote: ] This interpretation is suggested, we think, by John of St. Thomas in Phil. nat. I, q.26 a.2 (Reiser, II, 529A12); IV, q.10 a.22 (Reiser, III, 308A and 312B). Though these texts offer useful directives for research, they do not contain a satisfactory definition of an 'objective instrument.' In the form in which John of St. Thomas presents it, this notion retains unresolved obscurities."
Yves R. Simon, An Introduction to the Metaphysics of Knowledge, trans. Vukan Kuick and Richard J. Thompson (New York: Fordham University Press, 1990), 124-126.