Two tests of all-or-none learning and retention

A number of recent experiments have seemed to imply that the basic associations of learning are formed in an all-or-none manner. However, the evidence provided by these experiments has been essentially indirect since the presence of an association has been assessed by an all-or-none method, viz. simple recall. By allowing the subject more than one attempt at recall or recognition more direct evidence can be obtained. All-or- none learning implies that the probability of success during further attempts will be at the chance level. The experiments described show that this is not the case for the stimulus materials used. On the other hand the uncertainty in bits during further attempts was a fairly high proportion of the uncertainty expected on the all-or-none hypothesis. For this reason the results may indicate only a minor breakdown of the hypothesis. In a retention test administered after a week, fewer first-attempt successes were achieved than in the immediate test but, despite this, the uncertainty after a first- attempt failure was about the same as in the immediate test. The experiments also incorporated a second test of the all-or-none hypothesis. This involved comparing 2-choice and 4-choice recognition scores and recognition scores with recall scores. After corrections for guessing, these scores should be identical, if the hypothesis is correct. Results of this corrected-comparisons test tended to support those of the further-attempts test.     

Verbal learning: Incremental vs. all-or-none theory

The incremental and all-or-none theories of verbal learning are compared by means of a little-used but methodologically superior variation of the “drop-out” paradigm with paired associates. Earlier experiments purporting to be relevant to the controversy are rejected as failing to offer a conclusive distinction between the two theories. The results presented here are taken to support the incremental theory. It is suggested that irregularities in the results of this and other experiments are caused by several intrusive factors, and a “dual-factor” hypothesis which was put forward to account for these irregularities is questioned on logical and methodological grounds.     

Detection of repeated trigrams: evidence of all-or-none learning

Two experiments investigated trigram detection in a continuous recognition task. In Experiment 1 consonant trigrams were presented visually, one at a time, with occasional repetition of a trigram after an interval of 0, 2, 4, 8, or 16 other trigrams. Subjects were told to respond with a button press every time they saw a repeated trigram. If a subject responded to a repeated trigram, it was not repeated again. However, if a subject did not respond to a repeated trigram, it was repeated again at the same interval for up to 3 repetitions. For all intervals greater than 0, the probability of noticing a repeated trigram did not increase with the number of repetitions. In Experiment 2 meaningless shape trigrams were presented, and occasionally a trigram was repeated after an interval of 0, 1, or 2 trigrams. For both intervals greater than 0, the probability of noticing a repeated trigram did not increase with the number of repetitions. The results demonstrate that a repeated input does not necessarily leave a permanent trace in memory.     

Statistical methods for a general theory of all-or-none learning

Greeno and Steiner have shown that a three state Markov Chain with a single absorbing state is equivalent to many of the current formalizations of All-or-None learning theories. Distribution statistics and other summary statistics are derived from the general model. Expressions for the maximum likelihood estimators of its parameters and the sampling variances of the estimates are presented. Likelihood ratio tests for several different null hypotheses are derived. These tests permit one to evaluate the usual null hypotheses in terms of the parameters of a process model.This research was supported in part by a USPHS Predoctoral Research Fellowship, 1-F1-MH-31-289-01, by a grant from the Graduate School of the University of Texas, and by grant HD-02212-01 from the National Institute of Child Health and Human Development to Dr. John Theios.     

Individual differences and the all-or-none vs incremental learning controversy

This paper provides an analysis of the ability of the all-or-none and incremental learning frameworks augmented with individual difference assumptions to account for several standard all-or-none learning demonstrations. The paper allows the one element model (OEM) and the linear operator model (LOM) to have a completely arbitrary individual difference distribution. Among the results are: 1. the LOM can not account for either the Rock replacement studies or Bower's constancy of the error given error curve. 2. The OEM can account for Rock's and Bower's results only if there are no individual differences. 3. No independent responding scheme can jointly predict Rock's and Bower's results. 4. A beta distribution on the learning rate allows the LOM to achieve a close fit to at most one of Rock's and Bower's results. 5. An RT⁽ⁿ⁾ experiment can not rule out an independent responding scheme for any n, n ≥ 1.     

Markovian processes with identifiable states: General considerations and application to all-or-none learning

It often happens that a theory specifies some variables or states which cannot be identified completely in an experiment. When this happens, there are important questions as to whether the experiment is relevant to certain assumptions of the theory. Some of these questions are taken up in the present article, where a method is developed for describing the implications of a theory for an experiment. The method consists of constructing a second theory with all of its states identifiable in the outcome-space of the experiment. The method can be applied (i.e., an equivalent identifiable theory exists) whenever a theory specifies a probability function on the sample-space of possible outcomes of the experiment. An interesting relationship between lumpability of states and recurrent events plays an important role in the development of the identifiable theory. An identifiable theory of an experiment can be used to investigate relationships among different theories of the experiment. As an example, an identifiable theory of all-or-none learning is developed, and it is shown that a large class of all-or-none theories are equivalent for experiments in which a task is learned to a strict criterion.This research was supported in part by the National Science Foundation under grant GB-319.     

Evidence for all-or-none preperceptual processing in perceptual set

“Set” is known to improve the accuracy of report of the stimulus attributes for which Ss are set to respond. The set may produce this effect by acting on either preperceptual or postperceptual processes, or both. To elucidate the manner by which set produces its effect, this paper explores the consequences of two assumptions concerning set (1) On any given trial, a S is either “set” or “not set”: (2) if “set” by the stimulus information. S’s perception of the stimulus actually improves, whereas, if “not set” by the stimulus reformation, the S’s report correctness is improved by an independent interaction of stimulus and set information on a response process. Evidence is provided for both these assumptions. In addition. It is argued that only set given before stimulation can affect a perceptual process.     

The use of categorial organization: Not an all-or-none situation

Second and fifth graders memorized each of two lists of words that differed in category representativeness. In addition, they received one of three memory instructions: standard free recall, repetition, or categorial processing. Recall performance was about equal for standard vs repetition, and both were lower than category processing, especially with the highly representative items. Age did not enter this interaction, so it is concluded that at both age levels, the accessibility of categorial information influences children's use of it. This suggests multiple types of possible production deficiencies.     

Speed-accuracy decomposition yields a sudden insight into all-or-none information processing

The existence of discrete all-or-none information processing has often been assumed as a basis for stage models and also as an important characteristic of nonlinear connectionist models; however, there has been little or no hard empirical evidence supporting the existence of this phenomenon. In search of such evidence, we applied speed-accuracy decomposition (Meyer et al., 1988), a technique for detecting partial response information, to the examination of the time-course of processing in a (Gestalt) insight-like task, namely, anagram solution. This task was chosen because the Gestalt psychologists conjectured that insight is a sudden, discrete phenomenon. Supporting this view, we found little or no evidence of partial information in two experiments, thereby providing what may be the strongest evidence to date for all-or-none processing.     

A theoretical and empirical comparison of the all-or-none multilevel theory and the mixed model

Two theories for stimulus generalization in paired-associate learning (P.A.L.) are compared—the mixed model and the all-or-none multilevel theory. Both theories postulate basic all-or-none learning processes but differ in their mechanism for generalization. The mixed model postulates that generalization occurs only during responding; whereas, the multilevel theory postulates that generalization takes place during learning itself.     

Consciousness isn’t all-or-none: Evidence for partial awareness during the attentional blink

Alternative views of the nature of consciousness posit that awareness of an object is either an all-or-none phenomenon or that awareness can be partial, occurring independently for different levels of representation. The all-or-none hypothesis predicts that when one feature of an object is identified, all other features should be consciously accessible. The partial awareness hypothesis predicts that one feature may reach consciousness while others do not. These competing predictions were tested in two experiments that presented two targets within a central stream of letters. We used the attentional blink evoked by the first target to assess consciousness for two different features of the second target. The results provide evidence that there can be a severe impairment in conscious access to one feature even when another feature is accurately reported. This behavioral evidence supports the partial awareness hypothesis, showing that consciousness of different features of the same object can be dissociated.     

Intentional fixation of behavioural learning,or how R-O learning blocks S-R learning

Four experiments deal with the acquisition of knowledge for the control of voluntary behaviour. Subjects had to accomplish a computer-controlled learning task that required them to learn which one of four actions (R) had to be performed in order to attain a certain one of four outcomes (O) in the presence of one of four situational contexts (S). Reinforcements were assigned to actions according to two schedules: In the S-R condition each of the four actions was consistently reinforced in the presence of a certain situation, whereas in the R-O condition each of the actions was either always (Experiments 1-3) or mostly (Experiment 4) reinforced if a certain outcome was required. Furthermore, the type of feedback was varied. In Experiment 1, only positive or negative reinforcements were fed back, whereas in Experiments 2-4 the actions resulted in outcomes that had to be compared with the required outcomes in order to determine successes and failures. The results indicate a preference for learning R-O contingencies over learning S-R contingencies. Most subjects were so fixated on learning R-O relations that they remained completely blind to the consistent reinforcement of S-R mappings. The data suggest that, in line with the ideomotor principle, the acquisition of behavioural competence is based primarily on the formation of bidirectional action-outcome relations. Specifications of the underlying learning mechanisms are discussed.     

Implicit visual learning and the expression of learning

Although the existence of implicit motor learning is now widely accepted, the findings concerning perceptual implicit learning are ambiguous. Some researchers have observed perceptual learning whereas other authors have not. The review of the literature provides different reasons to explain this ambiguous picture, such as differences in the underlying learning processes, selective attention, or differences in the difficulty to express this knowledge. In three experiments, we investigated implicit visual learning within the original serial reaction time task. We used different response devices (keyboard vs. mouse) in order to manipulate selective attention towards response dimensions. Results showed that visual and motor sequence learning differed in terms of RT-benefits, but not in terms of the amount of knowledge assessed after training. Furthermore, visual sequence learning was modulated by selective attention. However, the findings of all three experiments suggest that selective attention did not alter implicit but rather explicit learning processes.     

Effects of learning contexts on implicit and explicit learning

Two parallel tasks involving rule learning were identified in Experiment 1A and were used to assess implicit and explicit learning. In both tasks, subjects had to input numbers in order to reach the target values of outputs. The relationship between inputs and outputs was either simple (in the simple task) or complex (in the complex task), and the way in which target values were presented could be in the form of either numbers (in the simple task) or lines (in the complex task). Experiment 1B examined the validity of the explicit measure in the complex task. Experiments 2–4 investigated the interaction between learning contexts and the simple/complex learning tasks. Verbalization and instructions to search for the rules facilitated the simple-task learning and hurt or have no effect on the complex-task learning. In the observational-learning condition, no learning occurred for the simple task, and the complex task learning was impaired. These results suggest that the complex task and simple task involve two distinct learning systems. Other implications are also discussed.