Implicit Learning

by MarkMcKergow

Scientists in unconscious learning breakthrough shock horror probe....

Amongst the latest publications, "The Handbook of Implicit Learning" (Stadler & French, 1997) caught my eye. Described as "a readable, rigorous overview of one of the most exciting domains in contemporary cognitive psychology; the role of non-conscious processes in thought, affect and behaviour.", I was intrigued to find out what the academic psychology community is now investigating.

Unconscious learning so far

In the world of SEAL, the idea of unconscious learning is nothing new. In the 1950s, Georgi Lozanov's ground-breaking experiments in Bulgaria showed that phenomenal rates of learning and retention could be achieved, in part by using para-conscious and unconscious methods as well as more conventional conscious approaches. By using suggestion, music, art and stories, the learning is introduced at several levels simultaneously (see for example Lozanov, 1979 and Dhority, 1991).

Norman Dixon's classic and controversial work in the 1970s went even further, pointing out that stimuli which were completely beyond conscious awareness could be responsible for learning (Dixon, 1971 & 1981). Paul Scheele has built on this work in Photoreading, where defocussing of the eyes is used to allow unconscious input of complete pages of a book, which is then used in further processes of review and learning (Scheele, 1993). A good brief review of unconscious learning processes can be found in Scheele (1997).

The "Implicit Learning" described in this weighty tome (636 pages!) appears to have surfaced as a field in the last 5 years. Papers are being published at an increasing rate (about 20 from 1993 - 1995) but there has been no single work attempting to pull the whole thing together until now. The Handbook is very much aimed at an academic audience, and tends to the obtuse in terms of language. I have summarised what follows from the vast array of data to give a flavour of the subject.

Defining Implicit Learning

How do these scientists define implicit learning? This is a crucial point, as many of the researchers have somewhat different ideas, and is discussed at some length. Basically, there are three different parts to the psychological learning process which could be implicit or explicit:

Perception: Implicit learning is defined here as those cases where perception is explicit, as distinct from what is called subliminal learning, where inputs are below the perception threshold.

Learning: Whether the "things to be learned" are made explicit or not. Here, they are kept implicit, ie not stated in advance.

Retrieval: Whether the learning can be recalled or used at will (explicit), or reappears only in certain contexts and/or without conscious effort from the subject (implicit). Both of these possibilities are allowed under the heading "implicit learning".

Investigating Implicit Learning: Sequence Learning

The results of several kinds of experiments are presented in the Handbook. For brevity, I will describe just two different types here: sequence learning and artificial grammar tests.

In sequence learning experiments, subjects are asked to respond to a sequence of tones or visual signals by pushing buttons. The process is repeated up to 11 times, with a repeating sequence and random sequences used; in one example, the first 7 sequences are the same, then 2 random sequences, followed by 2 more goes at the initial sequence. The subjects are not told of this plan, are simply told how to respond to individual tones, and are further distracted by being given a tone counting task to occupy their conscious minds still further. Their reaction times are measured, and are deemed to be indicative of learning - shorter reaction times imply a "learning" of the sequence. [Fig 1]

The results of such an experiment are presented as a graph of reaction times against sequence number; a typical example appears as Figure 1. (The Handbook is crammed with such graphs!) The graph shows response times speeding up as the sequence is repeated, and getting slower when the accustomed sequence is replaced by a random one. This type of experiment measures implicit learning with implicit retrieval - no attempt is made to have the subjects actually say what the repeated sequence is, they simply respond faster when it arrives.

Artificial Grammars

In artificial grammar experiments, subjects are required to divine some rules underlying the construction of strings of letters. Some strings are "valid" in the experiment (ie are in accordance with the rules) whilst others are not. In a classic series of experiments in the 1960s, George Miller and Noam Chomsky initiated "Project Grammerama", an effort to understand how people went about the learning of artificial grammars and thereby shed light on the ways we learn real human grammars. These experiments ran into brick walls - they were dull to participate in, and only the simplest artificial grammars could be used before participants got bored and gave up.

In a revisiting of these experiments, Robert Mathews and Barbara Cochran have found ways to make such experiments more productive. Instead of having subjects deal with meaningless character strings, they invented a story about labels on cans of food, some of which (those whose labels were valid strings) were said to have been poisoned. Subjects then had the task to work out how to predict whether a certain label indicated poison, and to generate a list of as many poisoned labels as possible.

The results were indeed a lot more useful. Attempts to generate valid strings by a trial-and-error process was again frustrating, with a hit rate of only 5.5%. Groups who were given examples of valid strings, and then asked to identify other valid strings, did much better (30% to 40%). This confirmed the hypothesis that grammar is better learned by exposure to instances, rather than trying to figure out the rules. This has an interesting resonance with Pinker (1995), who hypothesises that the human brain is created with a "language instinct", a predisposition for unconsciously identifying and following language rules exemplified by children's efforts at rendering regularity into language ("I runned" for I ran, or "I'm the winnest!" for I'm the winner!).

A SEAL perspective

These experiments are fascinating when viewed from our perspective as people interested in engendering real learning in real people of real and meaningful things. It is excellent to see the idea of implicit learning being taken seriously by the scientific community. The ideas around directing the learner's attention to one thing whilst giving examples of other things at different levels is a key (and in my view under-emphasised) part of accelerated learning.

One way in which the researchers may care to extend their attention is to be careful about the meaning and importance of the tasks to the subjects. We know that learning which has meaning is more easily taken up than that without. In the tone-sequencing tasks above, the experimenters are careful to avoid implying meaning, presumably to keep their experiment "clean". I suggest that they may be throwing the baby out with the bathwater - having people "learn" meaningless sequences may be rather like trying to make a cake with only flour and water - halfway there but doomed to very limited success. Similarly in the artificial grammar experiments, a modest level of meaning helped the process. What would happen, I wonder, if the subjects were given a series of more meaningful symbols (flags or colours, for instance?) and then told that finding all the valid strings would provide them with a dream holiday or a big donation to a charity of their choice?

Using implicit learning

There are various ways in which we can incorporate aspects of implicit learning into our everyday work in the classroom or training centre.

1. Use structures at the level of process to model content - for example, teaching coaching skills by using coaching techniques, or having maths students learn about geometry whilst placing themselves in different arrangements in the classroom while they work.

2. Using peripherals like posters and objects - and not referring to them. My experience is that learners can remember the contents of poster in the training room - as long as they are moved about regularly and stand out with colour etc.

3. The importance of para-conscious behaviours by the teacher/trainer - conveying far more of their beliefs about the learner's ability in the ways they talk and stand than by what they actually say. Take care to build self-esteem and have positive expectations of the learners, whatever labels they come with.

4. Communicating with metaphors and stories seems to work at an implicit level - suggestions can be made and parallels established, leading to significant change in the learner for no "apparent" reason.

5. Some implicit learning can be made explicit by "unpacking" the experience afterwards. At other times leaving it implicit is fine - especially if the behaviour or knowledge can be seen to have been installed. Indeed, sometimes revealing the "way it's done" (for example pointing out the metaphor that made the difference) can give the learners reason to doubt the efficacy of what they've achieved!
 

References
Dhority, Lynn: "The ACT Approach", Gordon & Breach (1991)
Dixon, Norman: "Subliminal Processing, The Nature of a Controversy", McGraw-Hill (1971)
Dixon, Norman: "Preconscious Processing", Wiley (1981)
Frensch, Peter: "One Concept, Multiple Meanings", in Stadler and Frensch op cit. (1997)
Lozanov, Georgi: "Suggestology and Outlines of Suggestopedia", Gordon & Breach, (1979)
Pinker, Stephen: "The Language Instinct", HarperCollins (1995)
Scheele, Paul R: "The Photoreading Whole Mind System", Learning Strategies Corp (1993)
Scheele, Paul R: "Natural Brilliance", Learning Strategies Corp (1997)
Stadler, Michael and Frensch, Peter: "Handbook of Implicit Learning" Sage (1997)

Mark McKergow advises organisations, trains trainers and designs learning events using implicit learning and other accelerated learning methods.