Stochastic designs and stochastic learning objectives

We need more accurate terminology to describe the instructional strategies we employ in our designs. The term learning objective is far too vague to be of much use. This blog post explains how the term fails and offers an additional term, stochastic learning objective, to clarify differences among instructional strategies. Not all learning is stochastic, but much is.

Detail from Jackson Pollock’s No. 8, where the artist harnesses the stochastic nature of drips to accomplish an esthetic. The mantra is create the context, the product will come.

Disciplines within education use the same term in different ways, confusing everybody and wasting time. In instructional design (IDT) programs, objectives suggest discreet performances that are either mastered or not. There is no flexibility. Mastery is obvious. Scholars point to Ill-defined as a design term suggesting content that may have multiple correct associated performances, but the term confounds content and performance. Language Education also uses the term objectives, albeit in a different way. Language educators are not talking about ill-defined content. They have a pretty solid idea of correct language performances, even if they choose not to predict those performances. In Language Education, designers of instruction will not have confusion about the content and still see a range of mastery performances; objectives often revolve around communicative tasks. Discrete linguistic performances do not determine mastery of, or even failure to master, an objective. Rather, Language Education has evolved the term, objective, to be broader and more appreciative of multiple communicative strategies. The definition of the term is less flexible in other areas of education. Study abroad programs and immersion programs are great examples of the stochastic instructional strategy. In such cases, discrete language performances may not determine success or failure of a study abroad participant, but by virtue of the experience, learning is likely to have occurred. The instructional design of the study abroad experience endeavors to create a context where learning is likely to take place. Language immersion programs work much the same way.

Stochastic Learning Objectives are a set of competencies or knowledge items that we hope learners will come to in the process of learning, but our instruction is not explicitly targeted on the performance itself. Often, there is no single discrete performance that constitutes mastery. Rather, the instructional design is aimed at building a context where the learning of skills is likely to occur, and multiple performances, taken holistically in or around the educational context, indicate learning. The terms stochastic learning, and stochastic learning objectives, are useful for conveying the point others have already made.

Stochastic designs are everywhere outside of online learning

The relationship between instructional strategies and stochastic learning has not been labeled in the past, but the concept has been around a long time. Noam Chomsky told the students and faculty at Arizona State University that if we knew what the outcomes were before we started teaching, it would not be education. Education, he says, is joint discovery that happens among teachers and learners. When we start, we don’t know the answer, and that’s the objective— to learn together. He stops short of drawing a distinction between education and instruction, but his logical trajectory is not hard to guess. In contrast, Mager has made a career of explaining instructional performance objectives. In Mager’s objectives, the answer must be known ahead of time; otherwise, it is not instruction at all. Standards and school districts spend countless hours debating objectives. A lot of confusion arises because we don’t have a common meaning to the term. We draw distinctions in isolation from each other. Stochastic learning objectives offers a way to identify instructional strategies that do not hinge on single items, but rather rely on probabilities.

In stochastic designs, the probability that one will learn, or not learn, is often separated from the sequence of instructional events. Instructional sequences need not be dictated in stochastic designs, nor are learning items necessarily contingent upon one another. In a stochastic design, a mastery performance need not come together in a  specific way. A parallel example exists in visual art. Jackson Pollack is often credited for capitalizing on the stochastic nature of dripping paint– at some point the process, the artist will come upon the aesthetically pleasing image (see public domain image in this post). His admirers’ credit him not for dripping paint particularly well, but rather for knowing when to stop. Recognizing when the performance was complete.

PhD programs are also stochastic in many ways. An example of a skill that is typically approached stochastically is scholarly hedging. Early researchers are cautioned not to make claims prematurely. In this process of learning, they go from having ideas to having hedged notions. In lay speech we find the term “idea” quite often, but when we switch to academic discourse, ideas seem to be replaced with notions. Notions are less rigid than ideas; it’s ok for a notion to be vague. The notion is a wonderful example because while I find that graduates of PhD programs have notions and others have ideas, I have never heard of any course or advisory bullet point directed at teaching PhD students to have notions and not ideas. In fact, it is ludicrous to think of it. Stochastic outcomes are quite common in discourse learning. I have had that notion for the past few years; previously I only had ideas. 😊

Not enough online learning uses stochastic designs

At-a-distance learners rarely encounter stochastic designs because we plan the stochastic nature out of our learning interventions by disassembling content too much. Online learning emphasizes pin-point precision, and carefully supported performances where the undisputable correct performance is obvious at every stage. That is far easier to accomplish when a task is dissected into constituent parts. Bite-size content with clear performance objectives accompanied by lots of support via examples and supplemental material are common recommendations for at-a-distance designs. Evaluation systems such as Quality Matters celebrate these highly targeted, discrete point, overly scaffolded designs. We’ve developed a system that purposefully avoids stochastic designs even when it is exactly what learners need. Removing all of the stochastic nature out of designs disempowers learners, and weakens their ability to think and explore.

Removing all the stochastic elements within a design is a disservice to our learners. Instructional designs that lack an element of exploration will stop short of education, and get stuck in instruction. No amount of supplemental material will prepare learners for some of the learning they need; namely, the attitude and disposition to move up the high order thinking continuum. For example, I assigned learners to collaborate via a wiki. My goal was to support critical thinking about technology choices in online collaboration. A specific learner, through no fault of her own, experienced some of the less attractive attributes of wiki collaboration. Her work was overwritten because she had saved just moments prior to another student’s saving a different version of the wiki page. This effectively deleted her entire contribution. A second experience revolved around incompatibility issues with how the collaboration rendered on different devices. From a Quality Matters perspective, this instruction was poorly designed. However, in the context of graduate learning in instructional design, these events effectively brought about the learning I had hope to support. I had hoped to support a critical discussion of media affordances in light of asynchronous collaboration. had we approached the topic discussing each affordance separately, the discussion would have been dry, disengaging, and unhelpful.

In my stochastic intervention, I could not have predicted these events even though I suspected the collaboration might be clumsy. No number of forewarnings could have generated the deeper learning that took place when learners recognized the negative attributes of wikis. These must be experienced for the concept to be learned. As a teacher, I did not intend for students to delete one another’s work. Rather, I wanted learners to grasp both the positives and negatives of the tool. The learning objective was stochastic, and meant to empower, rather than to lead learners through a predetermined process. Realizing the larger ambitions for online learning will require recognizing where stochastic interventions fit in, and where they do not.