Tagline: An Essay on the Cognitive Unconscious

ISBN:0-19-505942-5

My rating: 83/100

See Book Notes for other books I have read. If you like my notes, go buy it!

Key Points and Important Quotes:

• If you’re reading this book you will want to become familiar with two methods of psychological testing. PL = Probability Learning and AG = Artificial Grammar
• Implicit learning is the acquisition of knowledge hat takes place largely independently of the conscious attempts to learn and largely in the absence of explicit knowledge about what was acquired.
• People appear to be, generally speaking, arational. … [they] did not seem to know what they knew nor what information it was that they had based their problem solving or decision making on.
• Virtually all interesting complex human skills are acquired beginning with the labored, conscious, and overtly controlled (declarative) processes of the novice that gradually give way to the smooth, unconscious, and covertly controlled (procedural) processes of the expert. [Consequentially, it seems important to me that we learn to love the process of being the beginner.]
• Passive observation can be a more effective teaching tool than active testing. Passive observation of event sequences (of say, red and green lights that flash alternately in a special pattern with certain rules) is just as effective and much faster (~15x!) than active training, asking the subject to guess the next light color for example. What does this mean? To use an example, if you are trying to learn French do not start by attempting to reproduce sentences you have learned – instead spend your time listening to more and varied content.
• When people are presented with an environment that is structured, they learn to exploit that structure.
• In artificial grammar experiments where the rules were complex, subjects that were informed that the grammar followed rules ended up creating false rules that didn’t match the actual grammar rules in the experiment. They did worse than subjects that had no instruction at all. The lesson: in complex environments, don’t make written “rules” to explain the behavior of the environment, instead rely on your intuition since it is much more powerful and reliable.
• Looking for rules won’t work if you cannot find them. The converse is also true: Looking for rules will work if you can find them.
• Most of our formal education is handled as though the acquisition of complex knowledge were an explicit, conscious process, despite the fact that we know that much of the critical knowledge that a child must bring to an educational setting was acquired implicitly.
• True understanding is not easily communicated. [Therefore, you don’t always have to be able to explain yourself in words to know something]

1. Introductory Remarks

Implicit learning is the acquisition of knowledge that takes place largely independently of the conscious attempts to learn and largely in the absence of explicit knowledge about what was acquired.

nativism vs. empiricism

nativism – in the mind from birth (Chomsky and Fodor)

empiricism – learned through the senses

One of the standard heuristics in evolutionary theory is that phylogenetically older and more primitive structures will display telltale classes of properties different from the more recently evolved. One of these is that the structures with greater antiquity tend to be more robust and resilient, less prone to disruption of function than the newer. Therefore, we would expect to see implicit cognitive processes show greater resistance to interference from neurological insult and clinical disorder than the explicit processes. … We should expect to find fewer individual differences between people when implicit processes are in use than when explicit processes are.

We would expect to find that the more primitive a function is shown to be, the more refractory to consciousness it will be.

α – symbol used to represent the unconscious

β – symbol used to represent the conscious

The relationships between instruction, performance, and intelligence will be better understood if the distinctions and interactions between implicit and explicit learning are taken into consideration.

… learning occurs independent of the learner’s awareness of the process.

… tacit knowledge, knowledge whose origins and essential epistemic contents were simply not part of one’s ordinary consciousness.

One of the unspoken (implicit?) elements of the period during which the early implicit learning work was being carried out was that humans are rational and logical and they reach conclusions and make decisions based on coherent patterns of reflection and analysis.  … It became increasingly apparent that people do not typically solve problems, make decisions, or reach conclusions using the kinds of standard, conscious, and rational processes that they were more-or-less assumed to be using. PP People appear to be, generally speaking, arational. … [they] did not seem to know what they knew nor what information it was that they had based their problem solving or decision making on.

lacunae – plural of lacuna – empty space or missing part

Hasher and Zacks argued that [encoding the frequency and location of objects and events in the environment] was a primitive and fundamental cognitive process and, as such, was relatively unaffected by variables such as age, developmental level, IQ, and affective state, which normally have considerable impact on cognitive processing.

Declarative knowledge – knowledge that we are aware of and can articulate

Procedural knowledge – knowledge that guides action and decision making but typically lies outside of the scope of the consciousness.

Anderson’s view is that virtually all interesting complex human skills are acquired in a characteristic fashion. They begin with the labored, conscious, and overtly controlled (declarative) processes of the novice that gradually give way to the smooth, unconscious, and covertly controlled (procedural) processes of the expert.

Implicit memory is taken to have been displayed whenever a subject evidences that there was a memorial residue of an earlier experience in the absence of any comparable phenomenological sense of the previous experience.

The unconscious, covert systems are now recognized to be robust in the face of a host of disorders and dysfunctions that seriously impair the operation of the conscious, overt systems.

Amnesia – inability to consciously recall and, in severe forms, to recognize stimuli previously presented.

scopolomie – aka hyoscine or Devil’s Breath – used to treat motion sickness and postoperative nausea and vomiting.

Abrams and Reber (1988) found that psychotics and chronic alcoholics who were incapable of discovering relatively simple letter-to-number rules that were presented as explicit problems nevertheless showed intact ability for implicit learning of the underlying structure of a complex artificial grammar.

2. Implicit cognition: the data base

the position I take throughout is that of the primacy of the implicit. As will be argued below implicit learning is the default mode for the acquisition of complex information about the environment.

Our conceptual framework:

1. That the processes of implicit induction are general and universal.
2. That implicit learning is a foundation process that operates as a base for the development of tacit knowledge of various kinds.
3. That there are no a priori reasons for making assumptions about biological determinants of specific kinds.
4. That the properties of the tacit knowledge base that develops from implicit learning are reflective of the structure inherent in the stimulus environment.
5. That implicit acquisition is the default mode and the one normally adopted.
6. That when procedures are modified, functional considerations will dictate the degree to which implicit and explicit processes will be recruited.

Taken together, this cluster of assumptions represents a fairly strong form of classical representational realism.

There are fairly good reasons for concluding that the structre that is out there “in the world” is the structure that ends up “in the head” (see Reber 1991).

Experimental Procedures

Research in implicit learning is properly carried out using arbitrary stimulus domains with complex, rule-governed, idiosyncratic structures. Ideally, in order to obtain insight into a process such as implicit learning, it is essential to present the learner with a stimulus domain that has the following properties:

1. The stimuli need o be novel, the rule system that underlies them cannot be something that is already within the subjects’ sphere of knowledge – this, of course, is an important heuristic first used over a century ago by Ebbinghaus.
2. The rule system that characterizes the stimulus domain needs to be complex. If subjects are able to “crack the code,” as it were, by simply testing of consciously held hypotheses, implicit learning will not be seen.
3. The stimuli should be meaningless and emotionally neutral, or as devoid of meaning and affect as one can make them. The point here is to diminish extraneous aspects, particularly aspects that might have differential effects on individual subjects. [This will result in testing something that is not real. People do not do things unless they perceive them to be meaningful in some capacity, and removing the meaning from actions would necessarily remove the incentive to learn them. I think this assumption for an experiment might be useful but only if one is aware of this fact and accepts that experiment results are flawed.]
4. The stimuli should be synthetic and arbitrary. If our assumptions about implicit learning are correct, it should appear when learning about virtually any structured stimulus domain and the use of the synthetic and the arbitrary gives additional force to the argument.

We have chosen to work with two procedures that we have found to be extremely useful: artificial-grammar learning and probability learning (PL).

Probability Learning

Experimenters had two (or more) buttons from which the subject had to choose and, of course, two (or more) corresponding outcome events.

The probability learning procedure quickly became a classic. It contained all of the essentials for an associative acquisition process to occur: a well-defined stimulus that set the stage for a response, a fixed number (usually two) of unambiguous responses among which the subject had to choose (denoted A1 and A2), and a similar number of reinforcing outcome events (E1 and E2) – the response that occurred specified, and presumably reinforced, the corresponding response.

If there were two events, E1 and E2, and they occurred with probabilities of, say, .80 and .20 respectively, then subjects’ average response rates for the two responses, A1 and A2, over trials, typically came to approximate .80 and .20. Hence, the subjects displayed probability learning.

A number of difficulties with the standard model emerged.

As the length of a run of a particular event increased, subjects showed an increasing tendency to predict the other event. It was as though they were saying, “the right light has come on so many times in a row now that the left is due.” Jarvik (1951) called this effect negative recency: it is also known as the gambler’s fallacy. #cognitivebias

My note: Could one create a training/game using historical stock market data to condition an implicit learning response that beats the market? Set the investment time frame, feed the subject 1000 to 10,000 scenarios and condition the responses. I think the difficulty would be determining which information to use as the training data. I would start by finding the metrics or pieces of information that have the highest correlation to positive results in the market.]

What has passed in the literature for probability learning is actually a much more subtle process in which subjects learn implicitly about the stochastic structure of an event sequence to which they have been exposed.

Accordingly, the procedure was modified as follows. The subject begins an experimental session simply by observing the occurrence of a sequence of rapidly presented events. There is no ready signal and the subject makes no prediction responses. In this situation, a passively observed event turns out to be functionally equivalent to a traditional trial and learning session consisting of two or three minutes of observing events at a rate of two per second [~360 events total] is sufficient to put a subject at an asymptotic rate of responding. That is, subsequent prediction responses made by subject who have had this learning experience show all of the characteristics of ordinary subjects who have had an equivalent number of traditional trials. We dubbed this procedure the instant asymptote technique.

it seemed mildly counterintuitive that 3 minutes of observation could have the same impact on subsequent behavior as 45 to 60 minutes of ordinary probability learning training.

Empirical Studies of Implicit Learning

When people are presented with an environment that is structured, they learn to exploit that structure. That is, they come to use their knowledge of the structure to behave in a relevant fashion in its presence.

My note: This concept is far more profound than I originally thought. Could this be said differently; for say, the common criminal who has learned to exploit their environment to their own benefit? Perhaps an even greater argument for expedient and precise judgements.

Grammar Learning Studies

In this experiment subjects were purposely not informed that they were working with rule-governed stimuli and subjects were requested to memorize strings of letters printed out on cards, four strings to a card.

In this study maximum string length was 8.

Table 2.1 An example of the stimuli used in a typical artificial grammar study. The 20 Learning Stimuli are used in the acquisition phase, usually by asking subjects to memorize them. The Testing Stimuli are used in the well-formedness talk where subjects are asked to determine the grammatical status of each string when presented.

Learning Stimuli	Testing Stimuli
PVPXVPS TSSXXVPS TSXS PVV TSSXXVV PTVPXVV TXXVPXV PTTVV TSXXTVPS TXXTVPS …	PTTTVPVS* PVTVV* TSSXXVSS* TTVV* PTTTTVPS PVV PTTPS* TXXTTVPS TSXXTTVV PVSPVXPX* XXSVT* TSSXXTVV TXS TXXVX* … Indicates a non-grammatical string.

In several experiments where the rules were complex and not particularly salient, the capacity to articulate rules was found to be negatively associated with actual performance.

Matrix Scanning Studies

In the Lewicki et al. (1987) study subjects were confronted with a complex matrix of numbers arranged in four quadrants on a computer screen. The matrix contained a single target number whose location changed from trial to trial. The subjects’ task was to note the location of the target number by pressing, as quickly as possible, a button indicating the proper quadrant. PP The key trial was determined by the sequence of locations that it had appeared on the first, third, fourth, and sixth trials of that run (the locations on the second and fifth trials were selected at random). For example, if the target appeared in Quadrants 3, 1, 4, 2 on the critical trials, it would always occur in Quadrant 1 on the seventh trial.

On Implicit Versus Explicit Processes

The experiments reviewed to this point were all run under instructional sets in which subjects were not informed that the stimuli were structured or rule defined.

An obvious question is what effect would explicit instructions have? What happens when subjects are informed, at the outset, that the materials they will be working with reflect regularities and patterns?

The information about the event frequencies had virtually no effect on behavior. [Informed and uninformed individuals had the same results]

Postexperimental debriefings were revealing. … Subjects reported achieving a sense of the nature of the event sequence from experience with events that they did not derive from the explicit instructions. … Encouraging one group of subjects to search for the structure in the stimuli while a comparable group was run under a neutral instructional set (Reber 1976). Both groups were given the same learning phase, during which they had to memorize exemplars from a synthtic grammar and an identical testing phase during which they were asked to assess the well-formedness of novel letter strings. Unlike the earlier procedure, in this study informed subjects were told only about the existence of structure, nothing was said about the nature of that structure. PP The explicitly instructed subjects in this study performed more poorly in all aspects of the experiment that those given neutral instructions. They took longer to memorize the exemplars, they were poorer at determining well-formedness of test strings, and they showed evidence of having induced rules that were not representative of the grammar in use. … Explicit processing of complex materials has a decided disadvantage relative to implicit processing.

The simplest conclusion here seems to be the right one: looking for rules won’t work if you cannot find them.

My note: be wary of systemetizers.

… the converse of the earlier conclusion: Looking for rules will work if you can find them.

The Effects of Providing Explicit Information

They were presented with Figure 2.1 [a diagram of rules for an artificial grammar]

The manner of interaction between these two modes of apprehension was explored by introducing the explicit training at different points in the observation period. One group of subjects received the explicit instruction at the outset before any exemplars were seen; one group received it partway through the observation period; and a third group had the explication of structure delayed until after the full set of exemplars had been observed. As in the typical grammar learning study, knowledge acquired during learning was assessed by a well formedness task. PP the earlier during the observation training the explicit instructions were given, the more effective they were. PP The most plausible interpretation here, and the one that has interesting implications for theories of instruction, is that the function of providing explicit instructions at the outset is to direct and focus the subject’s attention.

hagiographic – the writing of the lives of the saints; a biography of a saint

… present the “wrong” one to a subject and the instructions will not have a salutary effect.

My note: A reason to seek out experts and expert information when beginning to learn a new, complex task.

pg 55 my note on Fig. 2.8: Essentially, changing words in a language but keeping the grammar the same is easy for us to learn, but changing the grammar and keeping the words the same is more difficult to learn.

… the same subjects working with the same grammars can emerge from a learning session with either an instantiated memory system or an abstract one, depending on the learning procedures used.

emendation – an alteration intended to improve.

the implicitly acquired epistemic contents of mind are always richer and more sophisticated than that which can be explicated [described in words, or conscious].

α – the sum of information available to the unconscious

β – information available for conscious expression

3. Evolutionary considerations: the primacy of the implicit

otogeny – the origin and development of an individual organism from embryo to adult. 

phylogeny – the evolutionary development and history of a species or trait of a species or of a higher taxonomic grouping of organisms. 

Darwin’s selectionist model has had a greater impact on psychological theory, and research than any comparable theory.

propadeutic or propaedeutic – providing introductory instruction.

Nativism – Chomskyan or Fodorian

its opposite is Behaviorism – of the Skinnerian type

VON BAER’S LAWS. In 1828, Karl Ernst von Baer, the German embryologist, put forward four laws of [evolutionary] development. … The essence of von Baer’s arguments was that the embryos of higher species dd not and could not possibly pass through the adult forms of lower species as the recapitulationists maintained. The four laws express this rather succinctly:

1. The general features of a large group of animals appear earlier in the embryo than the special features.
2. Less general characters are developed from the most general, and so forth, until finally the most specialized appear.
3. Each embryo of a given species, instead of passing through the stages of other animals, departs more and more from them.
4. Fundamentally therefore, the embryo of a higher animal is never like the adult of a lower animal but only like its embryo.

These laws are now captured through the generalization that differentiation proceeds from the general to the particular.

developmental lock model – The notion of a developmental lock emerges from considerations of the hierarchical manner in which complex systems are constructed or evolve.

Schank and Wimsatt (1987) show that a number of important general principles of evolutionary biology are proper theorems of the formalization of the developmental-lock model – including von Baer’s laws. The following are seven theorems from the 1987 paper are relevant to the thesis that I am developing here.

1. Evolution is increasingly conservative at earlier stages of development; features that emerge at earlier developmental stages change at lower rates.
2. Most adaptive evolution takes place through modifications that occur late in development.
3. Features expressed early in development have higher probability of being required for those expressed later.
4. Differentiation proceeds from the general to the particular.
5. In general, features expressed earlier in development are evolutionarily older.
6. Features that have persisted for a significant time tend to become ever more resistant to change as more and more features become dependent on them.
7. Comparative, cross-species analyses can generate significant information about the structure of developmental programs.

Some generalizations

It is possible to distill the above propositions of von Baer and Wimsatt down to four, much more general principles, as follows:

1. The Principle of Success. Once successful forms emerge, they become the foundation for later forms. All of evolution is testimony to this principle. It also is a simple entailment of the developmental lock model.
2. The Principle of Conservation. Following from the above, developmental processes are conservative. Once successful forms are established, they tend to become fixed and serve as foundations for emerging forms. For example, once the set of jaw bones in early fish became modified and began to function to transmit vibratory information, the form of the inner ear was fixed. Virtually every species that has an auditory sense uses these three ancient bones in this fashion. This principle follows directly from both von Baer’s first two laws and Schank and Wimsatt’s first, second, third, and fourth theorems.
3. The Principle of Stability. Following from the preceding, earlier appearing, successful, and well-maintained forms and structures will tend toward stability, showing fewer successful variations than later appearing forms. Under some circumstances, variations are selected for and under other conditions they are not. Given that the environment remains stable, it becomes relatively unlikely that variations that occur in earlier forms will prove adaptive. Hence, the relatively fixed, successful forms will display greater stability than the later appearing forms. The stability principle is reflected by the cladistic classification systems in evolutionary biology, which are based on there having existed an ancestral species from which the members of the clade are descended. This principle is embodied in Schank and Wimsatt’s fifth and sixth theorems.
4. The Principle of Commonality. Evolutionarily earlier forms and functions will be displayed across species. This principle, like the others, follows naturally from the preceding. Von Baer’s fourth law generalized to the phylogenetic level clearly entails it, and it is directly embodied in Schank and Wimsatt’s seventh theorem. It also lies at the heart of much of ethology and comparative psychology and, of course, is a virtual axiom of behaviorism.

cladistics – a system of classification based on the presumed phylogenetic relationships and evolutionary history of groups of organisms. 

In order to link these principles up with the material reviewed in the first two (or three) chapters and put the cognitive unconscious into an evolutionist context, we only need one more step: A simple axiom.

An Axiom About Consciousness

Consciousness is a late arrival on the evolutionary scene. Sophisticated unconscious perceptual and cognitive functions preceded its emergence by a considerable margin.

Hypothesized Characteristics of Implicit Systems [in light of the evolutionary framework]

1. Robustness: Implicit learning and implicit memories should be robust in the face of disorders and dysfunctions that compromise explicit learning and explicit memory.
2. Age independence: As compared with explicit learning, implicit acquisition processes should show few effects of age and developmental level.
3. Low variability: The capacity to acquire knowledge implicitly should show little in the way of individual-to-individual variation. Population variances should be much smaller when implicit processes are measured than when explicit processes are.
4. IQ independence: Unlike explicit processes, implicit tasks should show little concordance with measures of “intelligence” assessed by standard psychometric instruments such as the commonly used IQ tests.
5. Commonality of process: The underlying processes of implicit learning should show cross-species commonality.

parafoveal – (or parafovea; or parafoveal belt) – a region of the retina that circumscribes the fovea. Fovea – (or fovea centralis) – a tiny depression in the center of the macula, which is at te center of the retina. Gives the sharpest vision.

Rovee-Collier and her colleagues have presented evidence that young children quickly[, implicitly,] pick up the relationship between their own motor movements and the environmental impacts that they have.

Most of our formal education is handled as though the acquisition of complex knowledge were an explicit, conscious process, despite the fact that we know that much of the critical knowledge that a child must bring to an educational setting was acquired implicitly.

Ceci and Liker (1986a, 1986b) and Wagner and Sternberg (1985, 1986) have both reported cases where there is little relationship between scores on standard IQ tests and ability to function in complex, real-world settings.

4. Implicit issues: some extensions and some speculations

My summary on pg 119: The scientists running the experiments made up rules to govern things like artificial grammar and mistakenly thought their subjects would learn those rules. In reality, the subjects made up their own “rules” to explain the stimuli that didn’t match the rules used to generate the stimuli, but could make excellent predictions of the environment nonetheless.

“It’s neurons all the way down.”

PA – Paired Associate

We are stuck here in one of those classic muddles in psychology. The organism, and its attendant processes, is apparently so flexible that it avails itself of a variety of different processing modes, each sensitive to the constraints of the task and the demands of particular settings. The question that remains unanswered is what would be the default mode, if there were such a thing? That is, “all things being equal” what is the mode of functioning that we would anticipate finding? There are no clear answers, only further speculation.

The giant sea slug, Aplysia californica, with precious few neurons and barely anything that can be generously called a brain, clearly has the capacity for differential conditioning. … Aplysia is processing information and coding events.

A normal, intact frog has a fairly impressive range of behaviors … Leg twitches accompany stroking along the frog’s side, hopping follows a pinch of the hindquarters, tongue flickers are elicited by flying objects of appropriate size and flight patterns, and the like. If one removes the higher brain centers of the frog [it changes] virtually the entire range of behaviors are still present; strokes still elicit leg twitches, pinches hopping, and flies tongue extensions. What seems to be missing is not the raw behaviors, which are in large part mechanistic, reflexive, and wedded to particular lower brain structures, but the volitional elements, the operant aspects of action. The careful observer quickly notes that the frog seems to have lost its essential frogness.

Consciousness is likely emergent in even fairly primitive amphibian forms. Let’s call this form Consciousness I.

Consciousness II feels introspectively very different from the more primitive form, and there is precious little evidence that it is found in species other than the higher primates.

Subjects know they know something. 

True Consciousness II is characterized by both self reflection and the capacity to use the knowledge derived from self-reflection to modulate other functions, to have a causal role to play in other cognition.

There are good reasons for thinking that it was [Freud’s] dismissal of the traditionalist’s view of consciousness that was even more disturbing to his intellectual contemporaries [than his preoccupation with sex and other “base” motives].

Prediction and generation of events

In this section I want to deal with might seem, at first, to be a silly question, namely, “If you know what’s going to happen, does that mean that you know what’s going to happen?”

limen – a threshold below which a stimulus is not perceived or is not distinguished from another.

Neoteny is the “retention of juvenile characters by adult descendants produced by retardation of somatic development” (Gould, 1977b); mutations are the saltatory changes in genetic material brought about by factors other than Medelian recombinations.

Polydactylism – the presence of supernumerary fingers or toes.

True understanding is not easily communicated.

Studies on implicit learning suggest that school curricula should be modified to include more exposure to the variations that the specific subject matter displays and less energy and time should be spent on specific tutoring of rules and formulas. … Maximal learning takes place when there is some direction provided at the outset about the underlying nature of the environment.