As a comment on those items highlighted by both Mark Zollinhofer and David Jonassen, I would like to suggest it may not be valuable to look too deeply into learning styles but rather to concentrate upon learning approaches. That is, to consider in particular the characteristics of deep learning and surface learning and the cognitive strategies associated with both approaches. The problem with learning styles is that there are as many styles of learning as there are learners: learning style is largely a function of personality together with variables that are so diverse they are not useful as a means of analysing cognitive strategies employed by learners. (They might be of some use in "study skills" learning and are intrinsically of interest--but they can serve little value in providing analytical frameworks.)
Further, it is evident that learners change their approaches to learning (i.e., either deep or surface) to a large extent, according to task perception--that is, learners can learn to use cognitive strategies that are characteristic of deep learning approaches. This is where the debate on evaluation associated with task is so important. Also of interest are findings to suggest that surface learning strategies can be used as part of a deep learning approach.
This notion of learning approach is only one part of a larger pictureÑthe learning system. For more on this see: Biggs, J. (1993). Approaches to Learning. New York: Prentice Hall.
A big question for instructional designers (and this is where I agree with the original Jonassen paper so much) is explaining just why it is that so little deep learning occurs despite the systematic delivery of both knowledge and of techniques for manipulating it, even when they are carefully designed to achieve instructional goals. One answer is that knowledge cannot be "delivered" in this sense at all; knowledge is not "acquired"--it is constructed. This is why we must draw efforts away from instructional design and put them into provision of what Jonassen calls "unintelligent" tools--cognitive tools. (For various perspectives on this, see Olsson's (1993) paper in the Journal of Computer Assisted Learning, 9, 194-221; and the reply to that paper by Mayes in the same journal).
Also, a comment on cognitive tools. It seems a topic of the moment--especially with the notion of performance support (cognitive) tools/systems. I would like to a make a more considered comment on this topic a little later (if and when time permits)--but for now would like to draw attention to the wealth of work that has been done in this area in the UK by those who saw a place for knowledge construction tools back in the mid 80's. Nichol, Briggs, Conlon, Yazdani, Brough, Cumming, Mental Models Group (London)--the list is a long one. What is important is the work completed on the use of knowledge construction tools (cognitive toolkits) built in Prolog that allowed learners to become what was coined "knowledge engineers."
And can I further suggest that constructivism seems to have been used (misused?) in this debate as a "politically correct" paradigm--it is useful but it is only one. And, it is not a matter of a simple dichotomy between objectivism and constructivism as epistemologies--there is not a continuum connecting these two supposed extremes. There are many dimensions in learning theory, not all fit along an imaginary continuum--this is where Tom Reeves' work on evaluation of instructional technologies is questionable (Reeves, T. (in press) Evaluating what really matters in Computer Based Education). A continuum suggests that something or someone can be placed along its linear measure, that it, or they, can lean more or less towards either constructivism or objectivism--this is surely a misrepresentation. Constructivism offers a theory of knowledge acquisition that shares characteristics with other cognitivist theories and is different to objectivismÑbut it is not on the same continuum.