__________________________________________________________________________ | COPYRIGHT NOTICE: | | | | You may forward this document to anyone you think might be interested. | | | | The only limitations are: | | A) You must copy this document IN ITS ENTIRETY, WITHOUT MODIFICATIONS, | | including this copyright notice. | | You do NOT have permission to change the contents or make extracts. | | B) You do NOT have permission to copy this document for commercial | | purposes. | | | | The contents of this document are copyright (c) 1977 by the Journal | | of the American Society for Psychical Research. | | | | It was posted on the University of California at Davis ftp server by | | permission of the copyright holder. This ftp server contains ASCII | | files of published articles by Professor Charles T. Tart. Individuals | | wishing to obtain other documents there (which are added to from time | | to time) should | | Connect to ftp server, "ftp.ucdavis.edu". | | Log in as username "anonymous". Send your e-mail address | | as the ident/password string. | | cd to /pub/fztart. | | A "dir" command will show you what is available. | | A "get" command will retrieve documents. | | The file "currentcontents" will be updated regularly, showing | | what papers are available, perhaps with an abstract of each. | |__________________________________________________________________________| Toward Conscious Control of Psi Through Immediate Feedback Training: Some Considerations of Internal Processes Charles T. Tart (Note 1) Originally published with the above title in the Journal of the American Society for Psychical Research, 1977, Vol. 71, pp. 375-407. note------------ Note 1: My thanks to the est Foundation and the Parapsychology Foundation for their generous support of my research. end of note------ ABSTRACT: To achieve conscious control of psi, a motivated, active participant in a repeated-calling psi test must learn to scan and categorize the contents of his experiential field on each trial before making a response. If immediate feedback as to the correct target is then provided, he can form hypotheses as to which aspects of his experiential field (operating signals) are associated with correct psi responses and what sorts of control strategies are likely to produce useful operating signals. Immediate feedback allows testing and refinement of these hypotheses. The common decline effect in psi performance is seen as caused by the confusion resulting from lack of immediate feedback in most earlier studies. Conscious control of psi would eventually involve recognition by the person of momentary operating signals associated with success, such that he would go ahead and respond, and/or recognition of operating signals (or lack of them) associated with failure, such that he would not respond, but would wait for or induce a change in his internal condition that was associated with success. Various factors affecting immediate feedback training are discussed. INTRODUCTION Some years ago (Tart, 1966)1 argued that the standard para- psychological procedure of having participants in psi experiments make repeated calls at a multiple-choice target, such as cards, with- out the provision of feedback until after the completion of many trials, constitutes an extinction paradigm. In almost all conventional learning situations some kind of feedback on performance is provided almost immediately after each response, and under these circumstances we usually see an improvement in performance with practice. Eliminating this immediate feedback is a standard way to induce extinction, a steady worsening of performance with repeated practice. Thus I argued that the commonness of the decline effect in psi experiments, where scoring ability is eventually lost with re- peated testing, is just what we would expect. The provision of immediate feedback, on the other hand, should allow some partici- pants in an experiment to improve their performance. Some colleagues who read my original theoretical article did not grasp the importance of the prediction stating that if immediate feedback were given, only some persons would show learning. Rather, they mistakenly read it as predicting that all individuals, or groups of individuals as a sample of a population, would show an increase in scoring. When a percipient (Note 2) is calling a limited set of targets such as five geometrical figures, he will be right by chance alone approximately one-fifth of the time. Since the model of internal processes underlying my original learning theory, delineated only briefly in the original 1966 publication, was that immediate feedback allows the percipient to compare aspects of his ongoing experience with success or failure in responding, the presence of chance-correct responses is an impor- tant confusing factor. note----- Note 2: Throughout this paper I shall use the term percipient where conventional usage would lead the reader to expect subject. The term subject has acquired too many connotations in the psychological (and, unfortunately, the parapsychological) literature of a person who is manipulated, deceived, and subjected to an experimental treatment or manipulation. The undesirable effects of this, in terms of the creation of unrecognized, covert variables that distort experimental results, are just beginning to be fully recognized (see, e.g., Silverman, 1977; Tart, 1977a). The older term percipient is used here with the implications of a person functioning with the experimenter in a social situation of mutual trust and respect, with both genuinely motivated to learn psi, and without covert manipulation.' The arguments in this paper are about such percipients, and cannot be applied to a subject in a typical experimental situation without considerable modification. end note---------- The percipient should learn what kinds of experiential cues or patterns of cues are associated with success so that he can make his call when he recognizes the presence of that momentary state. He should also learn what kinds of cues or patterns of cues are associ- ated with failure so that he can either pass (i.e., not make a call), wait for that set of cues to change, or deliberately try to alter his momentary internal condition to one that has been associated with success cues in the past. Because he is sometimes right by chance alone, this means that he associates particular sets of internal condi- tions with being right, but these sets of conditions have nothing to do with learning to use psi. The presence of chance-correct successes, then, constitutes a kind of noise in the procedure or, more precisely, they constitute an inherent extinction procedure built into any repeated-calling situation. We shall consider the nature of extinction in detail later. The balance between the confusion/extinction in- duced by chance-correct responses and psi-influenced correct re- sponses will be an important determinant of whether learning takes place for a given percipient. I originally expressed this point in the concept of the talent threshold: if a percipient was above some minimal, threshold level of psi talent when he began feedback training, the proportion of correct responses due to psi should be high enough to permit learning to predominate over the extinction inherent in any repeated-guessing situation. If a percipient had some psi ability, but was below this talent threshold, there would be a shifting balance between a possi- ble learning process and an extinction process; here we would gen- erally expect to find a slowing down of decline or extinction, usually to the point of giving relatively stable psi performance for short- to moderate-length experiments where fatigue, boredom, loss of moti- vation, and the like did not become important factors. For a percip- ient with no detectable psi to begin with, the provision of immediate feedback would, in general, have no effect, although if enough such percipients were tested there would undoubtedly be some with hidden psi abilities that might flower in this situation. These considerations led to another prediction made in the origi- nal publication-namely, that the degree to which a percipient prof- its from the provision of immediate feedback will be directly propor- tional to the degree of demonstrable psi ability he had to begin with. I believe that these three basic predictions-namely, that given some psi talent to begin with (a) the provision of immediate feedback will stabilize performance in most percipients, (b) will allow learning in some percipients, and (c) the degree of learning will be propor- tional to the incoming psi talent level of the percipient s-have now received strong support. An initial review of the literature and evi- dence from my first extensive training series have been presented elsewhere (Tart, 1975a, 1976). In the original theoretical article (Tart, 1966) and the publications mentioned above, I deliberately kept theorizing about the internal psychological processes involved in learning psi performance to a minimum, preferring to emphasize the importance of immediate feedback from a conventional, behavioral point of view. The pur- pose of this article is to elaborate my theoretical modeling of the psychological processes involved in learning to use psi under condi- tions of immediate feedback. (Note 3) Empirical support for the theory is discussed elsewhere (Tart, l975a, 1976; Tart, Palmer, and Re- dington, submitted for publication). While I shall focus on ESP, this discussion is readily applicable to other aspects of psi, such as psychokinesis. note-------- Note 3: I wish to acknowledge helpful comments on a preliminary version of this model (that I circulated privately in 1969) from Gardner Murphy, Karlis Osis, and Gertrude Schmeidler. The current version of this model has been further refined through my work with John Palmer. end note------ BACKGROUND: OUR ORDINARY STATE OF CONSCIOUSNESS I do not consider learning to use psi more effectively to be an easy task, even for the highly talented percipient. A brief look at the nature of our ordinary state of consciousness, which is the state in which we usually try to test or use psi ability, will provide back- ground for this contention. This understanding of our ordinary state of consciousness is spelled out in much greater detail elsewhere (Tart, 1974, 1975b). By virtue of being born human beings, we have a vast range of potentialities that could be developed, including the potential for using various kinds of psi abilities. I do not know if the potential for using psi is universal, but I believe it is widespread. What is impor- tant, however, is that each of us is born into a particular culture, which is a group of people who, through historical processes, only recognize the existence of a limited number of the total spectrum of human potentialities, and so allow many of them either to remain latent or to become essentially unavailable by not being developed within a critical time period. Further, of those human potentials known, some are labeled "good" and receive active encouragement for development, while others are labeled "bad" and are actively inhibited: in our culture, psi potential is either unrecognized or is put in the latter category for most of us. Although there is some ambiva- lent recognition of things like "intuition," which might sometimes involve psi, there is almost no active encouragement of it in the enculturation process. The enculturation process, carried out through selective pressures and guidance, rewards and punishments, from parents, teachers and peers, takes our unorganized human potentialities of infancy and, in the course of growing up, organizes a much smaller number of these into our "normal" or ordinary state of consciousness, a state that might be more accurately characterized as "consensus consciousness' '-a habitual and quite restrictive pattern of mental functioning that reflects the consensus reality, the cultural reality of our particular society. The fully developed consensus consciousness represents a largely automatic, routinized, habitual set of psycholog- ical structures and operations that implicitly guides our perceptions, our fantasies, our evaluations, our feelings, and our actions into channels and patterns that are approved of by our society. Insofar as psi phenomena frequently have no valued place within our particular consensus reality, particularly in the educational establishment, the immense force of habit and conditioning makes it difficult for us to commit ourselves wholeheartedly to using or developing psi facul- ties. Thus when we ask a percipient in the laboratory to attempt to use or develop psi, we can usually get only a limited amount of attention focused on this task, and there is a tremendous amount of implicit mental habit militating against the appearance of any signifi- cant amount of psi. There are important individual exceptions to this, of course, but nevertheless it is the social reality that is a constant background to our work. Resistance to Psi Note also that the conditionings that limit our range of mental experience and behavior are not simply cognitive: they have very strong emotional components, such that if we seriously question~the explicit and implicit judgments and limits conditioned into us, we may become disoriented, experience generalized anxiety, and have specific guilt reactions. Making a conscious decision to try to use psi does not make this emotional conditioning go away, so we may have both conscious and unconscious emotional resistances to the idea of using psi. Again, there are enormous individual differences in the strength of these factors, but they are quite common in our culture. These cultural resistances to psi, which may be specifically ac- tivated by deliberately trying to use it, are not the only factors that deter people in our culture from attempts to develop psi abilities. Observation of our own mental processes will demonstrate that we do not experience a passive state of mind with specific cultural conditionings activated only when the requisite stimuli come in from the environment. To the contrary, our ordinary state of consensus consciousness is like a three-ring circus: we are constantly thinking, fantasizing, remembering, planning, reacting to our own thoughts and fantasies, etc. Our ordinary state of consensus consciousness is a dynamically active, interactive, exceptionally busy state of affairs. Our ability to pay deliberate attention is, as it were, almost totally used up by the continuously on-going thought and fantasy that is our common lot. Some of the appeal of trying to develop psi ability in altered states of consciousness lies in the fact that the sheer activity level may be reduced in some altered states, thus allowing more awareness to be focused on the psi task. But there is a very high noise level (mental activity irrelevant to psi functioning) in the mind of the typical percipient who comes to the laboratory to take part in a psi experiment. Further, this is a noise level over which most people have practically no voluntary control. Try not to think of anything at all for five minutes, and the point will be amply demon- strated. Again, as with the specific resistances discussed above, there are enormous individual differences, but most people in our culture have an extremely high noise level in their ordinary state of consciousness and generally can do very little, if anything, about it. Given this noisy background and granting the difficulty it creates in trying to hear the "still small voice" of psi, let us now turn to an overview of the psi process, as I understand it. OVERVIEW OF THE PSI PROCESS An overview of the relevant aspects of the parapsychological and psychological processes necessary to learn psi is modeled in Figure 1. A series of interacting psychological functions which I call the Sensing, Evaluation, and Decision Making system-the functions which allow us to detect various kinds of inputs, evaluate their meaning, and make some kind of decision-are central. This is where we find conscious awareness. The most prominent input to the Sensing, Evaluation, and Decision Making system, however, is the constant ongoing noise discussed above, the irrelevant thoughts, feelings, strategies for gratification, and 'fantasies, and our cognitive, emotional, and bodily attachments (resistances to altering or giving up) to these things that constitute so much of our consensus con- sciousness. ___ | T | | A | | R | | G |--------------------> Post-Response --->Sensory | E | Feedback Input | T | | |___| | Feedback | /|\ | | | | \|/ Noise | | Noise --- | | | | C |<___| | | | H | Response | | A | / \ | | N | | | | ____________ | N | _____________________|_|____________\|/______ | | | E | | |-->| Memory | | L | | Sensing, |<--|____________| |___| | Evaluation, | ____________ | | Decision | |Irrelevant | | | Making Noise |/--| thoughts, | | |_____________________________________________|\--| feelings, | | ! /!\ /|\ ! /!\ /|\ ! /!\ /|\ |strategies, | | ! ! | ! ! | ! ! | | fantasies, | | ! ! | ! ! | ! ! | |attachments | | _\!/__!___| _\!/__!_____|_ \!/__!___|_ |____________| | | Psi | | Psi-specific | |Unconscious| ------->| Receptor |-->| Information |-->| Processes | | | | Processing | | | |__________| |______________| |___________| ! indicates control and operating signal flow routes Fig. 1. Major known psychological and parapsychological processes involved in manifesting psi and learning to improve psi performance. Arrows represent information flow between various processes. The overall information flow process begins in the upper left-hand corner of Figure 1 with the target about which we wish to obtain information via psi. I assume that this information is continuously available from the target, "passes" through some currently un- known sort of channel, and arrives at whatever process or mech- anism serves as a Psi Receptor; the Psi Receptor then converts the psi information originating in the environment into the appropriate mental or neural signals within the percipient. I have also shown psi noise being carried over the Psi Channel: noise is defined in this context as any kind of psi information which, given the percipient's task to identify a specific target, is irrelevant to or interferes with this. Further characteristics of psi noise, includ- ing whether it is random or introduces systematic bias, have been discussed elsewhere (Kennedy and Taddonio, 1976; Tart, 1976, 1977c; White, 1976a, 1976b). We shall not consider aspects of the target or the channel further in this paper, but simply assume that psi information about targets is generally available. Intermittency Whatever the nature of the Psi Receptor (or Receptors) may be, it probably operates only intermittently in the vast majority of cases. More precisely, percipients respond in ways that manifest psi only intermittently, and while the intermittency might occur in the proc- esses labeled Psi-Specific Information Processing or in flirther Un- conscious Processes, in this modeling we shall assume that intermit- tency can occur independently in the Psi Receptor, in Psi-Specific Information Processing, and in relevant Unconscious Processes. Looking at this another way, most percipients are simply guessing most of the time; i.e., whatever determines their particular calls of the targets, there is no psi-relevant information detectably influencing them. This is the case for free-response psi tests as well as repeated-calling tests. I shall discuss guessing in detail below. Sometimes a percipient may be "hot," i.e., he may make a long string of psi-related responses. Considerable "psi-bursting" of this sort is apparent in data collected in my laboratory on learning to use psi, and will be treated in future publications. Information Flow Routes The psi information flow, shown by the heavy, solid arrows in Figure 1, may sometimes proceed directly from the Psi Receptor into awareness (as associated with the Sensing, Evaluation, and Decision Making system). Insofar as the process of psi reception may not have been very selective up to this point, i.e., there may be a lot of information coming in by psi other than that pertaining to the desired target, this direct information output from the Psi Receptor may be only partially useful: it contains irrelevant (psi noise) as well as relevant information. The percipient may or jnay not be able to discriminate the relevant aspects of this by means of the ordinary kinds of psychological functions available in the Sensing, Evalua- tion, and Decision Making system. An alternate route of psi information flow, perhaps operating instead of or sometimes simultaneously with the previous one, is from the Psi Receptor to the Psi-Specific Information Processing system. These are postulated mechanisms not used in ordinary sensory information processing, but which work specifically with psi information. Their function is to enhance the detectability or dis- criminatability of the relevant target information. Their internal functioning is ordinarily not accessible to awareness. An example of such a mechanism is one I have termed "trans-temporal inhibition" (Tart, 1977b). It involves psi-missing of immediate past and im- mediate future targets, a process analogous to lateral inhibition in ordinary sensory systems whereby (irrelevant) background activity of receptors laterally adjacent to the stimulated one is actively inhibited in order to sharpen perception. The output of such Psi- Specific Information Processing is shown as going into the Sensing, Evaluation, and Decision Making system, where it may receive further processing that we can be consciously aware of. We also know that psi information often undergoes processing by those aspects of the human mind we call the unconscious-aspects which, although not directly accessible to ordinary consciousness, show dynamic qualities of emotional significance to the percipient. I have shown psi information going from Psi-Specific Information Processing into Unconscious Processes,(Note 4) and the output of the un- conscious transformations of the information being presented to awareness in the Sensing, Evaluation, and Decision Making system. Psi information passing through Unconscious Processes is more likely to undergo transformations or distortions which fit in with a particular percipient's individual needs than information which comes directly from the Psi Receptor or the Psi-Specific Information Processing mechanisms. note-------- Note 4: It might also be the case that information can go directly from the Psi Receptor to Unconscious Processes without going through Psi-Specific Information Processing, endnote------ Operating Signals In addition to the psi information that is about the specific identity of the relevant target, I have shown in Figure 1 what I call "operat- ing signals" going from the Psi Receptor, Psi-Specific Information Processing, and/or Unconscious Processes systems into the Sens- ing, Evaluation, and Decision Making system. By "operating signal" I mean a quality or pattern of qualities (which may vary over time) which do not convey target i(1entity intormation per se, t)ut merely serve to indicate that a particular mechanism is operating. (Note 5) This might manifest as, e.g., a feeling of confidence, a feeling of "energy" in a particular part of the body, or a particular quality of imagery or mental functioning that can be distinguished from ordi- nary functioning. When one or more such operating signals are present in the experiential field, they can be used as an indicator that other contents of the experiential field which seem target-relevant may indeed be influenced by psi information or be psi information. Such operating signals are the basis for successful confidence call- ing. note--------- Note 5: There may be other kinds of operating signals that would tell a percipient about variations in the style of information processing, such as displacement effects, but, for simplicity, I shall only discuss operating signals of this "process on" or "process off" variety. endnote------ The Psi Receptor, Psi-Specific Information Processing, or Uncon- scious Processes systems may each have only one operating signal associated with its functioning, or a combination of operating signals. If more than one operating signal is perceived simulta- neously in the experiential field, a percipient could be even more certain that psi-related information was also present. If there are always operating signals created by the functioning of one or more of these mechanisms or processes (even if it takes a lot of training to detect them in the experiential field), then, in principle, a percipient could learn to use psi whenever the relevant mechanisms were operating and pass at other times. Here the ultimate level of psi functioning would be determined by the proportion of the time we could learn to stimulate these mechanisms (the control strategies, discussed below) to operate. If the operating signals are only inter- mittently or unreliably associated with the operation of these mech- anisms, or if they are weak compared to the noise level of the percipient's mind, then a lower inherent limit on the maximum level of psi functioning would result. Control Strategies I have also shown a "control signal" going from the percipient's Sensing, Evaluation, and Decision Making system to each of the three lower systems. In one or more ways, we deliberately attempt to make these necessary systems operate. Such control signals may take the form of merely wishing that psi will occur, or they may involve an elaborate behavioral or mental ritual that the percipient hopes will make psi manifest. The provision of immediate feedback ought to allow him not only to determine what sort of operating signals he should use as a basis for response decisions, but also to determine what sorts of control strategies reliably yield relevant psi information. He might believe, e.g., that "not trying" will facilitate psi; feedback training will allow him to compare such a strategy with a more active one. This is a basic model of psi-relevant information making its way into awareness, where it can be detected and evaluated so the percipient can decide how to respond. He should also try to inhibit the irrelevant thoughts, feelings, strategies, fantasies, and attach- ments which, as noise, flood the Sensing, Evaluation, and Decision Making system and impair its efficiency. Note also that sensory input coming into awareness constitutes noise since, by experimen- tal specification, any sensory input must be irrelevant to the task of detecting the target by psi. (Sensory noise reduction seems to par- tially underlie the success achieved with ganzfeld procedures; see, e.g., Honorton, 1974; Terry and Honorton, 1976.) One exception to this is the specific sensory input, following the overt response of the percipient, that brings in feedback information. Finally, storing strategies in memory and drawing such past memories back into awareness is extremely important, as we shall see when we consider the specific trial-by-trial process of learning to use psi. LEARNING TO USE PSI: THE FIRST TRIAL Let us assume that the percipient takes part in a multiple-choice psi test, such as guessing which of 10 unlit lamps on a panel in front of him has been randomly selected as a target. His task is to push a button beside the lamp which he decides is the target for his first trial, and the machinery is wired in such a fashion that he receives immediate feedback. The correct target lamp comes on as soon as he makes his response. He will know instantly if he was right or wrong and, if wrong, in exactly what way he was wrong, such as displacing close to the correct target. On this first trial, the percipient must consciously or semi- consciously decide on some kind of strategy for coming up with a response. This strategy may be a relatively articulated, conscious strategy, or a relatively unconscious strategy. He might, for exam- ple, decide to wait for a visual image of the correct target number to appear, or he might try to instruct some unconscious part of his mind to deliberately make an especially vivid image appear. He might decide to run his fingers around the circle of lamps to see if there is a special "feel" to the correct one. He might think about how the random target selection process works and what it might likely determine as a target, or he might call on his memories for ideas he has about what sorts of numbers generally "come up first." The Guessing Process: An important input to (or aspect of) the Sensing, Evaluation, and Decision Making system is what I shall call the guessing process. The percipient sits down at his psi task and is asked to produce a long series of responses in the absence of sensory information about the target. While we would like him to use psi, we can hardly expect him to do so on every trial. It is generally not socially acceptable in the experimental situation for a percipient to sit for long periods of time and give no responses at all; most of the time he is under pressure to just guess. The percipient consciously, or semi-consciously, starts a rela- tively mechanical and irrelevant process going in his mind that results in his experiencing a stream of images or ideas about possible targets, the digits one through 10 in our particular case. This is a relatively continuous reviewing of what the target alternatives are. In the absence of an idea of a target identity arising in some obvi- ously novel way (such as a vivid visual image), or of one of the ideas/images of potential targets being associated with a discrimina- ble experiential quality that might be an operating signal, the percip- ient can always respond with whatever call the guessing process is producing at a particular time. Deciding on a Response Whatever strategy he consciously or semi-consciously uses, the percipient finally gets some kind of output from it that he attends to and which leads him to make a particular response. This Sensing, Evaluation, and Decision Making process has been shown as the central block of Figure 1. Consider its inputs. The percipient has (irrelevant) guessing processes which affect him. He has all sorts of experiences that, in relation to the psi task, are noise, including the constant thinking and fantasizing along the consensus reality lines discussed earlier. Some of this activity is random noise, simply distraction from the task at hand, but some of it may amount to systematic noise that biases him away from suc- cessfully using psi. He may also have to deal with external noise from sensory input in terms of the effects of the experimental situa- tion, including experimenter biases. On the first trial, however, the Psi Receptor and subsequent, relevant information-processing mechanisms may also be functioning and producing operating signals as well as the target identity psi information. For the reasons discussed above, it might be a "still small voice" that is totally drowned out in the internal mental noise. But it might be heard, and might have some specific sorts" of experiential qualities, or operating signals, associated with it. The percipient may try to sort out and categorize this mass of ongoing experience and base his overt response on some quality (or pattern of qualities) of it that he hopes is relevant. Or he may try to ignore most or all of his experiential field and try to produce some form of control signal, engage in some sort of mental strategy that he hopes will make the target-relevant information clearly appear in his experiential field. He may combine these two strategies. He finally makes a decision and pushes a particular response button. Im- mediately the correct target is indicated and so he has feedback as to the correctness and usefulness of his first strategy. As we shall see, in order to consciously learn to use psi, he must attend to both this feedback information about correctness or incorrectness and to his immediately available memories of how he went about making this particular decision. He is still being subjected, in most cases, to considerable noise of the sort discussed above at this time. Making Use of the Feedback If the percipient wishes to profit from receiving feedback, he must now somehow categorize and remember in partial detail, or re- member in full detail, what his response strategy was on the first trial and its outcome in terms of success or (various degrees of) failure, and store this in his memory. This can become a highly complex process. Ideally, the percipient should store all information about his expe- rience at the time of making a decision as to what his response should be, as well as all feedback information, whether he was right or wrong and exactly how he was right or wrong in order to take partial successes (such as close spatial displacement, etc.) into ac- count. In reality, few persons in our culture have had practice in paying close attention to internal processes, so the percipient may have little or no awareness of those processes that led to his re- sponse. He may, to varying degrees, simplify the details and overall gestalt of his experience into just one or a few concepts that he stores, such as "Had a visual image of the correct target and I was successful." This conceptualization and memory process, then, hopefully includes relevant information about operating signals or control strategies for using psi, and probably also includes some noise, the irrelevant aspects of the total experience, that occurred either during the response process or the evaluation of the outcome. For simplicity I have assumed above that this was the first attempt to make a psi response, and that past experience of the task need not be considered. Let us also assume he is successful in calling the target on the first trial. THE SECOND TRIAL As we consider internal psychological processes on the percip- ient's second trial, past experience becomes quite important. He may have a deliberate response strategy as before: it may be the same response strategy he used on the first trial or he may decide to use a different one. Or he may just guess. Noise is again present, although it may not have the same configuration as before. Psi information may or may not actually be reaching awareness on this second trial: we cannot assume that the aspects of the mind which receive the psi information work on every trial, so on some trials there is nothing but noise to work with. But for simplicity, we will assume the Psi Receptor and Psi-Specific Information Processing systems are working on this trial. To make his decision on trial two, the percipient must again try to separate the psi signal from the noise; i.e., he must try to attend to and discriminate the psi signal and any identifying experiential char- acteristics (operating signals) it might have from all his other ongoing experiences. As a new and important part of this, he may consult his memory as to what he did on the first trial and the consequences of it. If his experience is somewhat similar he might repeat that pre- viously successful strategy, or he might feel that his momentary constellation of experiences has changed ~n a way that makes the previous strategy irrelevant. However he does it, he makes a deci- sion as to how to respond, makes an overt response, and again must attend to his decision-making processes, his response, and the con- sequences of this in terms of feedback as to correctness or incor- rectness. He must also evaluate how well whatever strategy he used seemed to work this time. Insofar as he used the previous strategy that was stored in memory, he must re-evaluate how well that previously successful strategy worked when it was applied in this second trial. Again, he would ideally store all of the experiential information about this trial (possible operating signals and control strategies) and the previous one, but in practice he will probably store only parts of the information available to him. LATER TRIALS Let us now skip ahead some number of trials and look at the internal psychological processes on trial N. This is like trial two except that the information which is now available to be consulted from memory is far more complex than it was earlier. Results (abstracted and schematized to varying degrees) of a variety of potential operating signals and control strategies and their successful and unsuccessful outcomes are stored (with varying degrees of fidel- ity) in memory. Some of them, if the percipient has been relatively successful, are relevant to the genuine use of psi. Some may seem relevant, but, as they were involved with chance-induced successes, are actually a form of noise. They are hypotheses about using psi that are false, but which have seemed validated so far because of coincidental association with chance-produced hits. To make his response now, the percipient must try to ignore (and/or try to lower) the irrelevant noise that is steadily going on in his mind and use an optimal response strategy to identify the signal which might come from the psi aspects of his mind. If he has been fairly successful up to now, he may continue to repeat whatever strategy he has used before. If he has not been successful, he may wish to use his memory of previous strategies in a negative way to try to devise an entirely different strategy. Or he may want to modify a previous successful strategy. For example, he might try to use a motor automatism for responding instead of waiting for a visual image to appear. He finally makes an overt response for trial N and receives ifri- mediate feedback. He must again attend to this information in order to store the consequences in memory, but by this time, there is already so much information stored in memory, and so much more information is going to come along to be stored, that he must, consciously or semi-consciously, start working out strategies to store information more efficiently. He simply can't carry everything in memory. How useful these simplifying schemes will be is an important question: he may work out an esthetically satisfying scheme, but it may not encode the really relevant portions of his experience. His long-term results with the immediate feedback should eventually enable him to evaluate his mnemonic strategies, assuming he has not become too confused or given up by this time, or become rigidly attached to nonadaptive response or mnemonic strategies. The Importance of Memory The need to store a great deal of information in memory in a useful form is an important part of the learning process, and should eventu- ally be the subject of much research. Given the fact that there is a limit to how much can be remembered, the percipient may need to figure out what kind of information he can stop storing in memory, or what kind of information already stored he can deliberately forget. He might make the decision, for example, to try to remember only trials on which he was successful and ignore the ones on which which he failed were as or more instructive than some of those on which he succeeded: knowing when not to respond is very condu- cive to effectively using psi. A great deal of internal experimentation must go on over the course of training for the percipient to work out a memory strategy that is optimal for him: individual differences are important here, but immediate feedback should allow any percipient to learn what uniquely does and doesn't work for him. Any factors which interfere with the efficient use of memory input, organization, or retrieval are detrimental to learning to use psi more effectively. The sheer load on memory is per se detrimental, so we would expect to see learning over relatively short, psycholog- ically homogeneous periods, such as a single run: such learning would usually fall off over a longer period, as memory becomes overloaded. Any distraction or interruption could be detrimental. Certainly the time intervals between training sessions would be detrimental, especially in the early stages of learning, when it would be difficult to hold a subtle, poorly learned memory strategy over the time span and insulate it from other psychological processes and real-world events which might disrupt or confuse it. As adaptive psi strategies become more consolidated, however, they should be bet- ter able to withstand the disruptions between sessions, with an eventual overall increase in learning from session to session, as well as within sessions. A promising line of research on the question of learning to use psi will center around techniques for helping percipients to clarify, organize, and retain their memories. Immediacy of Feedback The importance of storing multiple aspects of the experiential field in memory in order to devise discriminative and control strategies, as well as storing these strategies, is the reason why immediate feedback is so important. The longer the interval between the var- ious internal events that are scanned and the feedback about success or lack of it, the greater the opportunity for relevant aspects of memory to lose sharpness or for intervening events (external or internal) to interfere with and confuse these memories. If psi- associated operating signals were very clear to begin with, they could be stored in memory in a distinctive fashion that is resistant to interference; but, especially in the beginning of training, the operat- ing signals and other psi-associated cues may be quite amorphous and subfle Ideally, a percipient would' "stop" his mental activity between his last scan of his experiential field and the feedback, so there was no intervening, potentially interfering activity. In such an ideal case feedback could be considerably delayed. In most cases, however, feedback should be given as rapidly as possible, using electronic devices which make it possible for feedback to be (in terms of human time) instantaneous. In many earlier studies a percipient's responses were scored against the targets at the end of a complete run, but such feedback is probably too delayed to be of much value. How does a percipient remember exactly how he felt on the 11th trial, which was a success, versus the 12th trial, which was a failure? The only possible value end-of-run feedback may have is either as a motivator (if the per- cipient did well on that run), or perhaps as an indicator of a fairly long-lasting psi-conducive state, an aspect of learning discussed in a later section. Interference from the Guessing Process A percipient's guessing process is not a good random number generator in a statistical sense. It is usually a very biased source. While the targets in a well-conducted experiment are generated in such a way that each is equally probable and each is independent of the previous targets, the percipient's guessing process will probably have definite biases that over-generate some calls and under- generate others, as well as definite sequential dependencies. An example of the latter is the well-known habit of not calling enough doublets (repeating the same call twice in a row) when people are asked to generate random numbers. (Note 6) note------ Note 6: This nonrandomicity of call generation by percipients can be turned into an advantage, however, as it offers a possibility of separating responses produced by the guessing process from responses produced as a result of other mental processes. The psychological literature on response bias should be of value here. end note---- Cognitive/emotional/habitual attachment to the output of the guessing process is a major obstacle to using and learning psi. Suppose a percipient has guessed a 9 on the previous trial, and he has a very strong habit of guessing a 1 after a 9. His task is to attend to his experiential field for any contents that might be psi related, but if the psychological pressure to respond with a 1 is strong, it can keep him from adequately attending to his experiential field and so reduce his chance of using psi. Stanford's (1967) work on response bias is also relevant here. In a situation where we have immediate feedback, another prob- lem is created by the guessing process. The guessing process is probably not "free-running" in most percipients, i.e., uninfluenced by knowledge about previous targets, but is, to varying degrees, influenced by such knowledge. Although we design our target- randomizing process to have serial independence, and although we may inform the percipients of this, many (consciously or uncon- sciously) spend some time trying to "figure out the random number generator"; i.e., they keep track of what past targets have been and modify their guessing accordingly. A percipient is likely to believe, e.g., that target doublets are extremely unlikely. Thus if a S was the previous target, he may modify his guessing process to call almost any number but a 5. If the current target is a 5, however, this guessing strategy will make it very difficult for him to pick up any psi-related aspects of his experiential field that could inform him that the target is a 5: he has too strong a bias against it. Elsewhere (Tart, 1977b; Tart, Palmer, and Redington, submitted for publication), I have shown how a measure of this kind of maladaptive guessing process, which I call strategy boundness, is indeed significantly and negatively related to present-time hitting, trans-temporal inhibition, and short-term learning. Note that the number of alternatives in a guessing task will have some effect on the extent to which a percipient is bound by maladap- tive guessing processes. If there are only a few choices, say four targets, it is relatively easy and thus more tempting to try to keep track of which targets have occurred on previous trials and so try to "figure out the random number generator." If there are many choices, as in our 10-choice task, keeping track is more difficult, and thus less likely to be engaged in for long. The extreme of this dimension is a free-response task, where the well-nigh infinite number of possible alternatives on every trial dis- courages guessing processes based on keeping track of previous trials. This may be an important reason for the high level of success often found with free-response psi tests. Ideally, percipients should never initiate these sorts of maladap- tive guessing processes, but pay attention only to their current experiential field and deliberately stored strategies on each trial, and make a deliberate, conscious "estimate" of what the target may be. But, since I assume that the Psi Receptor and other relevant infor- mation processing mechanisms that pass the information into awareness may not be functioning at all on many trials, the percip- ient's "estimate" can then be based only on noise, so he will likely respond from his guessing process. If he can inhibit the guessing process on trials when his experiential field may contain psi-relevant information, or learn to pass when he senses no operating signals indicating psi-relevant information are present, that is fine, but the compelling nature of guessing processes and a need to make some kind of response, discussed above, can make this difficult. The provision of a "pass option" on any psi training device is probably helpful here, as the percipient can decline to make a call when he feels he has no real cues to the target. However, I have not yet investigated the usefulness of the pass option empirically. Holistic Processes I have talked about the process of storing relevant information and evaluation/response strategies in a linear, sequential kind of way, but it should be noted that we are not just linear, sequential beings but also holistic, pattern-sensing beings. We engage in the kind of sensing and thinking now popularly associated with right hemi- sphere functions of the brain, and we will eventually have to deal with these more holistic ways of sensing, storing, evaluating, and responding. SUCCESSFUL PSI PERFORMANCE Assuming a percipient whose Psi Receptor and Psi-Specific In- formation Processing mechanisms bring him psi information a fair amount of the time, and further assuming that there are experiential qualities (operating signals) associated with the operation of these parts of the mind that can eventually be discriminated, we can see now how conscious use of psi must be learned. Over a number of trials with immediate feedback, the percipient must gradually build up a set of conceptualizations about generation of control strategies and discrimination of operating signals, leading to various response strategies, and store them in memory. On any given trial he must then compare the components and pattern of his ongoing experience at that moment with the conceptualizations of discrimination and control strategies, and make one of several kinds of decisions. One decision might be that since the idea he has of the correct target is associated with experiential qualities that have been frequently associated with success in the past, he should respond with this particular idea of the target. A second decision might be that since the experienced conditions have been routinely associated with fail- ure, he should not make any response at the moment, but either wait for a new impression to arise that might be associated with experien- tial qualities related to success, or deliberately (Note 7) try to alter his momentary mental state to one related to success. note----- Note 7: In the course of training the percipient must, of course, note whether deliberately creating experiential conditions associated with success produces the same degree of success as spontaneously experiencing them. end note---- Altering Strategies These are not the only kinds of decisions a talented percipient might have to make. He might, for example, have worked out a strategy that produces a fair degree of psi success, but continued repetition of that strategy fails to result in any further improvement in performance. Now he must decide whether to stick with the strategy that has produced results, modify it, or discard it altogether and experiment with new strategies that might lead to a higher level of performance. If the decision is to experiment, a percipient who has been responding at a high and rather steady rate may suddenly drop back to chance-level performance (or even psi-missing). We would expect to see great variability in his performance as the new strategies are tried out. There seems to be a clear parallel here with some of the findings in biofeedback research. A subject is asked to try to gain voluntary control over some bodily function for which he has at the beginning no clear experiential representation. By means of immediate feed- back he learns what value that bodily function takes from moment to moment, and then tries to search within his (noisy) experiential field for particular cues (operating signals) which are associated with manipulations (control signals) that are successful in affecting that bodily function. For some tasks, such as modifying brain waves, subjects frequently find a strategy which gives them some degree of success, but after sticking with that strategy for a while, they find they can't go very far with it, and so must abandon it in order to try new strategies. Thus performance curves in biofeedback frequently show runs with increasing levels of success, leading to relatively stable performance, and then a sudden loss of success and great variability for subsequent runs. Note that the ability to do well in a biofeedback task, while obviously implying some degree of deliberate, conscious control, does not necessarily mean that the person can verbalize the strategy to himself or others. A percipient could learn to use psi fairly well without necessarily being able to explain just what he does. STABILIZATION OF PSI SKILL Although one of my main interests in teaching psi ability is in pushing such learning to as high levels as it can go, it is important to recognize that people enjoy success and dislike risking failure. When they have found a successful strategy they are often hesitant to risk tampering with it in order to try something else. I would expect that as we study the process of learning with immediate feedback more extensively, we will find a number of percipients who hit upon some reasonably successful strategy and stabilize at fairly high levels of psi performance, but who, for various reasons, will not risk changing that strategy in order to try something that might (but might not) be more successful. If the successful strategy has included learning to discriminate some cues that indicate the psi process is working and these cues are experientially clear, or if the percipient has found a clearly suc- cessful control strategy, he might be able to maintain his level of psi performance without further immediate feedback training, or with only occasional refresher training. My understanding of the mental processes of some successful psychics reinforces this view: they originally acquired some sort of special psi ability under naturalistic conditions that seem to have involved some degree of feedback, they learned to identify certain cues (a special state of conscious- ness, a certain kind of bodily quality, etc.) that indicated when they were 'hot," or they adopted a successful ritual; and they then became so attached to the rewards resulting from this partially successful performance that they are not willing to tamper with it. They may actively resist attempts to explore the mechanisms of their success through fear of undermining their effectiveness. They are then able to keep this strategy psychologically isolated enough from their other mental processes to prevent it from being blurred by the noise of the rest of their experience. The functional isolation of a (partially) successful strategy can thus be highly important. If the strategy is not isolated, then other kinds of experiences will be mixed with it, and it may become unclear just when the proper cues are present to act on an impres- sion that will turn out to be based on psi. This can explain why many traditional rituals for evoking psychic powers involve complex, but very stylized and rigid psychological processes: they are designed to produce a state of mental isolation around the particular psi function to keep it from coming in touch with ordinary mental processes that might confuse the issue. An example of temporary stabilization of psi ability in the course of immediate feedback training is provided by Kanthamani and Kelly (1 974b). Working with B. D., an exceptionally talented percip- ient, they reported that his performance in a single-card clair- voyance test using playing cards seemed to be superior when im- mediate feedback of results was, usually at B.D. '5 own request, withheld. Specifically, for 86 trials in which immediate feedback was withheld, B.D. scored 11 exact (suit and number) hits plus 8 number hits. These no-feedback trials were intermixed with feedback trials. For 122 feedback trials, he obtained only 5 exact hits and 13 number hits. A parallel finding was reported in an earlier study with B.D. (Kanthamani and Kelly, l974a). These results are illustrative of how successful immediate feedback learning of psi can be: after some feedback trials in which B.D. learned to develop a good strategy, he would know that he was "hot"; i.e., he could apparently detect operating signals associated with high success and so would deliber- ately request that feedback not be given on a forthcoming trial. Perhaps he wanted time to internally consolidate his mental state/ response strategies without the distraction of processing further feedback information, or needed the absence of distraction from feedback to encapsulate his successful states/strategies from inter- ference. THE NATURE OF EXTINCTION As a result of our discussion of learning, we can now look in detail at a phenomenon we have been taking for granted, extinction. In my original presentation (Tart, 1966) of the learning theory application, I used the term "extinction In a behavioristic sense, viz., a decre- ment in performance with continued practice, leading to disappear- ance of the response. A rat, for example, may have learned to press a lever when a green light is on, and this resulted in an immediate food pellet reward. We now change the situation so that pressing the lever when the green light is on no longer activates the food dis- penser. The rat still presses the lever frequently at first, then slows down, and finally stops pressing it altogether. We say the rat's lever-pressing behavior has been extinguished. We could just as well argue, however, that this extinction represents new learning: the rat gradually learns that there is no point in pressing the lever. The response contingencies have changed. The rat has not lost his ability to lever press. Suppose we have run a percipient through many psi trials without immediate feedback and find that his scores hold steady for a while, and may sometimes be increased or decreased as his motivation changes, etc., but then gradually decline until he is eventually scor- ing at chance expectation. Behavioristically this is extinction, but it is not the same phenomenon as with the rat, for the human perclp- ient seems to have lost his ability to use psi, even though we offer him inducements and rewards to manifest it again. What has hap- pened? I suggest that the percipient had some native psi talent to begin with that was occasionally activated under the conditions of the experiment, but that this ability had previously been functionally isolated from the rest of his mental processes, in the manner dis- cussed above. That is, his ordinary mental life rarely or never led to control strategies that might be (partially) successful in activating psi processes, and seldom brought much awareness of possible operat- ing signals associated with psi functioning. When a percipient first comes to a psi experiment, even one without feedback, his attention is repeatedly directed toward trying various kinds of control strategies and examining his experiential field because of natural curiosity and the task demands of the exper- iment. Gradually, however, this breaks down the functional isola- tion of his psi processes, and without immediate feedback there is no efficient way for him to constructively control this breakdown of subtle and delicate mental processes, nor to devise new and effec- tive control strategies and sense new operating signals to replace his original mental processes for manifesting psi. He gets confused. We get extinction through a confusion procedure. Depending on the degree of functional isolation of psi-relevant processes, the time for extinction will vary across individuals. In the extreme case of a percipient whose psi-relevant processes are very isolated, this can take a long time, especially if he uses them in a relatively automatic, unconscious way rather than subjecting them to conscious scrutiny. In this connection it is interesting to note the trend that developed along with the widespread use of card-guessing tests to deliberately try to keep percipients from being "self- conscious' '-a strategy that helped to maintain the functional isola- tion of the psi processes. Few people have any mental process existing in total isolation from all others, however, so eventually almost every talented percipient sooner or later became so confused about the delicate processes involved in demonstrating psi that he lost his ability to use them. Some declines, of course, have probably been due to other factors such as loss of motivation, so performance dropped to a chance level before the extinction process was very effective. LEVEL OF PSI TALENT AND LEARNING In any repeated-calling task, the percipient is right a certain pro- portion of the time by chance alone. As discussed earlier, these occasions constitute a kind of systematic noise which will confuse him; thus, an extinction procedure is built right into any repeated- calling task, even with the provision of immediate feedback. The ratio of psi-produced hits to chance-produced hits will determine the balance between learning and extinction processes. In the original presentation of my theory (Tart, 1966), I expressed this dilemma in two ways. One was the idea of a "talent threshold" above which learning would predominate over extinction. The other was the prediction that the degree of learning manifested under conditions of immediate feedback would be positively correlated with the degree of psi talent the percipient brought into the training situation. Let us consider these ideas in more detail. Figure 2 shows the kind of relationships I would predict for five dif- ferent categories of percipients. The vertical axis is the level of psi talent the percipient brings to the training situation (not the level he develops as a result of the training); the horizontal axis is a measure of learning or decline in the training, such as the slope of the performance curve over time. Two important assumptions underlie the model in Figure 2, namely (a) that whatever processes in a given percipient lead to psi performance are not totally isolated from other mental processes, so he will actively investigate and experiment with them, and (b) that each percipient is not only reasonably moti- vated to learn psi, but will stay reasonably motivated throughout the experiment. The second assumption thus restricts the following discussion to "short to moderate" length experiments that are not likely to lead to boredom and loss of motivation. "Short to moder- ate" needs to be determined empirically, although existing literature (reviewed in Tart, 1975a, 1976) suggests it may run as high as ten thousand or more trials. It will undoubtedly differ considerably across individuals. Psi Hitting | = Talent Threshold ____________________|____________________ | | | | | | | | | | | | Decrease --|--------------------|--------------------|-- Increase | | | | | | | | | | | | |____________________|____________________| | | Psi Missing [figure note: the shadings in the published figure cannot be reproduced in an ASCII file, but the text should make things clear] Fig. 2. Predicted relationships between the psi talent level percipients bring to immediate feedback training and the degree to which their psi performance will increase or decrease with training, for various talent level and motivation groups. Highly Talented Percipients The first category of percipients shown in the upper right corner of Figure 2 are those who are above the postulated talent threshold. (Note 8) These percipients show strong and individually significant psi-hitting prior to starting training. The theory predicts that the learning pro- cess will outweigh the inherent extinction due to chance-produced hits, so we would expect to see these high levels of psi associated with significantly positive measures of learning. While the relation- ship between psi talent level and measures of learning might be linear for part of the range for these percipients, it may be that above some high talent level learning becomes extremely rapid; if so, their regression line would flatten out toward the right of the figure as the speed of learning increased. note---- Note 8: I have deliberately put a break in the vertical hitting axis in Figure 2 to indicate that while the talent threshold lies above the level of individually significant hitting, we do not yet know its precise value, although I have estimated it to be around a psi-coefficient of .10 (Tart, 1975a, 1976). Note too that since learning is affected by motivation and other psychological factors, the talent threshold is a range of psi values rather than an absolutely fixed point. end note---- Moderately Talented Percipients The second category of percipients shown mainly to the upper right of the origin in Figure 2 is comprised of moderately talented psi hitters. Immediate feedback should stabilize performance for these percipients; that is, it should eliminate significant declines and stabilize psi-hitting at a moderate level. Variations caused by other psychological factors, such as motivation, do not allow for an exact dividing line between these percipients and the highly talented percip- ients, so Figure 2 shows an overlap here. An occasional moderately talented percipient might show a significantly positive learning measure, but, by and large, we would expect the data points of these percipients to be concentrated in the lower left corner of the upper right-hand quadrant of the scatterplot. Low or No Psi Talent Percipients In the third category I have combined percipients having low psi talent with those having no manifest psi talent. Those with no manifest psi talent should give a basically random, circular scat- terplot centered around the origin. I have shown this combined group skewed somewhat to the left (toward declines), however, as those with low psi talent would probably show extinction: the pre- ponderance of chance-produced hits would make the extinction process predominate over the possibility of stabilizing psi due to the immediate feedback. The degree of skew that needs to be put in here will require empirical study and specification of what "short to moderate" length experiments are: the longer the length of the experiment, the greater the likelihood of significant declines for the low talent percipients. The shorter the experiment, the more this combined group will tend to show a circular scatterplot, not reaching either significant psi scores or significant learning or decline meas- ures. For the three categories considered so far, the "rich get richer and the poor get poorer," with the "middle class" holding its own! Malfunctioning Psi Missers A fourth category of percipients can be distinguished, namely, low to moderately talented malfunctioning psi missers who make use of the immediate feedback to correct their malfunctioning and thus improve their scoring. These percipients do not have any emotional or cognitive need to "disprove" psi, as do those in the fifth cate- gory, so that they can profit from the feedback to correct the mal- functioning of their Psi Receptor, Psi-Specific Information Process- ing, or Unconscious Processes systems (see Figure 1) responsible for the distortion. Depending on the length of the experiment, their average psi level might be somewhat above, somewhat below, or even at chance since the early psi-missing can negate the later hitting. But they may have significantly positive measures of learn- ing as a result of going from missing to hitting. Thus, for an experi- ment of a given length we might have percipients who do not obtain significant psi scores, but nevertheless show significant learning. In a longer experiment, however, they might go on to show significant psi-hitting-assuming, as in all the above discussions, that they remain motivated. Motivated Psi Missers The fifth category, shown mainly in the lower left corner of Figure 2, is that of moderately talented, motivated psi missers who are, to varying degrees, statistically naive. These percipients have a con- scious or unconscious need to "prove" that psi doesn't exist, and have some psi talent operating in the service of this need. For the least statistically naive percipients, we would expect scores to clus- ter around the origin with neither significant hitting or missing, nor significant learning or decline measures. For the more naive moti- vated psi missers who interpret significantly negative psi scoring and/or a drop in scoring level as supporting their beliefs, we would see significantly negative hitting scores and significantly negative declines. There Is a sixth category of percipients that I have not shown in Figure 2 because it is not clear how to predict their performance. These are highly talented psi missers who have a strong motivation to "disprove" psi. While it would be tempting to put them in the left-hand corner of the lower left quadrant (Figure 2), I think that exceptionally strong psi-missing by such a percipient and/or obvious decrements in performance, when he had consciously agreed to participate in a study designed to produce psi-hitting and learning, would rapidly produce a conflict as he realized he couldn't get worse if there was no psi in the first place. What the outcome of this conflict would do to psi performance is not clear. Looking at the combined clusters of these various categories of percipients in Figure 2, it is clear that we would expect a roughly linear relationship between overall psi level brought into the training and our measure of learning. The highly talented psi hitters should contribute the most toward making this correlation significant, ig- noring possible nonlinearity induced in very high ranges. The mod- erately talented, statistically naive, motivated psi missers and the low psi talent percipients should also contribute toward bringing out a significant linear relationship by extending the regression line into the lower left quadrant. The percipients with no psi talent and the low to moderately talented malfunctioning psi missers who learned to hit would tend to create a great deal of variability around the middle of the scatterplot, thus tending to obscure the relationship between psi talent level and learning. This possible obscuration would be particularly serious in a study using percipients whose range of psi talent is limited, although in studies with large numbers of such percipients the relationship should still be statistically de- tectable. Interaction of Other Factors with Talent Level I want to stress again that the talent threshold is not an absolutely fixed entity. It is not the case that percipients with a low talent level can never hope to do more than stabilize their performance. Having a high talent level at the beginning of training will make learning much easier, but learning is affected by other things than just the frequency with which psi-related cues are present in the percipient's experiential field. One psychological factor which is likely to be related to learning is motivation. If a percipient is only mildly motivated to learn to improve his psi ability, he will not work hard enough at the demand- ing tasks of discriminating psi-related information, operating signals, successful control strategies, cutting down the noise from irrelevant mental processes, etc. A percipient who is strongly motivated to learn will work harder at these tasks and thus compensate for a lower psi talent level. We should, however, be aware of possible nonlinearities here: too strong a motivation, too strong an attach- ment to success, might introduce psychological difficulties of a different sort that would detract from the learning process. A second psychological factor which is also probably related to success in learning to improve psi functioning is general learning ability. We have psychological measures of general learning ability, but need more specific measures of the ability to (a) inhibit extrane- ous internal noise, to quiet one's mind; (b) ignore sensory input which might be distracting; (c) organize one's memory efficiently while building up a catalog of successful strategies; and (d) intro- spect and make fine discriminations among aspects of the experiential field. Other psychological factors will also probably turn out to be relevant. In initial research our prime goal will usually be to train a few percipients to show very high, reliable levels of psi so other aspects of the nature of psi may be investigated. As a practical measure to attain this end, it would seem most fruitful to concentrate our training efforts on moderately and highly talented percipients with high motivation, and make all experimental conditions, such as positive encouragement from the experimenter, as supportive as possible. MISCELLANEOUS CONSIDERATIONS Immediate Feedback as a Reinforcer In the psychological literature on learning, the term "immediate reinforcement" is frequently associated with the term "immediate feedback," and while they mean quite different things, it becomes easy to confuse the two. Immediate feedback, as discussed in this paper and my earlier writings (Tart, 1966, 1975a, 1976) on the appli- cation of learning theory to psi performance, is conceived as purely informational feedback to the percipient which he may make use of to modify his internal strategies which are relevant to learning and manifesting improved psi performance. This assumes, of course, that the percipient is motivated to improve his psi abilities. If he is not, he will probably pay little attention to the informational feed- back and derive little profit from it. Reinforcement refers to a process designed to elicit or strengthen motivation to perform on a task. If a hungry rat must learn to press two levers in a particular sequence in order for a dispenser to drop a food pellet, the appearance of the food pellet is not only informa- tional feedback, it is also reinforcement; it satisfies the rat's hunger need and so strengthens its motivation to perform the lever task properly. Similarly, in psi training feedback that the percipient is correct may, in addition to its informational function, serve to cogni- tively and emotionally reward him for attempting to use his psi abilities more effectively, and reinforce his longer-term motivation to continue trying to do so. Thus immediate feedback can also constitute immediate reward or immediate reinforcement, to varying degrees, for some percipients. There is clearly room for great variability here: if a percipient is already highly motivated to use his psi better when he begins the training procedure, the reinforcement aspect of the feedback will be relatively small. If his pre-existing motivation is low, or if it tends to weaken rapidly and needs repeated encouragement from the feed- back of being correct, then the reinforcement aspect of the feedback is much more important. The reinforcement aspect of immediate feedback involves an im- portant problem which should be explored empirically as well as theoretically: the optimal number of target choices to be used in a repeated-calling task. Based on the cognitive considerations dis- cussed earlier, as well as on my own experimental results to date, I believe that for percipients with high motivation a large number of choices is more effective in training psi because it keeps the noise level down: the proportion of times the percipient is right by chance alone is low in, e.g., a 10-choice task. At the other extreme, in a binary guessing task the percipient will be right for psi-irrelevant reasons 50 percent of the time, thus creating a very high confusion level. Some work of Schmidt's (1970) supports the idea that low p values are more efficient in eliciting psi. On the other hand, for percipients with low motivation to begin with or whose motivation needs to be strengthened frequently, a smaller number of target choices in the repeated-calling task might lead to better results by sustaining motivation. For any individual percipient, there might be an optimal number of choices for getting the best balance between strengthening motivation versus minimiz- ing confusion due to chance-induced hits. This factor needs empiri- cal exploration. My guess at the present time is that the confusion induced by many chance-induced hits is a detrimental factor which would not be outweighed by the strengthening of motivation pro- duced by increasing the reinforcement ratio, and I would prefer to use other means, such as (genuine) encouragement from the experi- menter, to keep motivation up in percipients who need such assist- ance. Also, if each percipient can choose the number of target choices for his training task, after some experience with different numbers, he might select the best balance on his own. Psi-Conducive Procedures Over decades of research a number of procedures have been discovered which, when successfully carried out, are generally (i.e., over groups of percipients) conducive to better psi manifestation. These include physical relaxation, friendly and relaxed experimenter-percipient relationships, reduced attention to sensory stimuli, hypnosis., ganzfeld stimulation, "waiting techniques" (White, 1964), game-like situations, and "don't try, just let it hap- pen" attitudes. While my theoretical approach is aiming toward deliberate, conscious control of psi, it is not contradictory to any of the above procedures, but complementary. While all of the above strategies may be helpful to some percip- ients, none are helpful to all. The point of providing immediate feedback of results is that it gives each percipient the opportunity to try one or more of these strategies, or strategies unique to himself, and develop or discard them on the basis of actual results. I would guess that a combination of hypnosis for quieting the noise of ordinary consciousness and producing increased attention to the experiential field, combined with White's "waiting tech- nique," would be particularly fruitful additions to immediate feed- back training. Momentary Versus Longer-Term Operating Signals I have discussed operating signals indicating that a useful psi- reception and/or information-processing mechanism is working on a particular trial in order to emphasize that the operation of these processes is intermittent and can vary rapidly. Both the experience of talented percipients and the fact that we often see fairly long bursts of psi functioning should remind us, however, that the appro- priate processes can also function for relatively long periods. We may see success on strings of consecutive trials, or a high (albeit not 100 percent) level of success over even longer periods. We can take these to represent the temporary existence of a "psi-conducive state," a constellation of psychological processes in the percipient that is relatively stable over a period of time, that is long compared with the length of the individual trial. Here we may have operating signals that indicate the presence of a psi-conducive state, rather than the momentary functioning of relevant psi information gather- ing and processing mechanisms per se. Immediate feedback training thus might allow some percipients to learn to recognize when they are in a psi-conducive state and respond rapidly, versus not being in such a state and waiting for one to appear or trying to induce it. Experimental Situation and Experimenter Effects It is important to recognize that a percipient neither manifests psi nor tries to learn to use it more effectively in a vacuum: he works in a particular experiment and social situation, set up by a particular experimenter or experimenters. As White (1976b) aptly put it, . it is difficult to see how the experimental situation can be separated from the experimenter, for in a sense it can be viewed as a trap which the experimenter has devised with the intention of catching a particular finding which will fulfill his hopes and expectations" (p. 334). Experimenter and situational influences can reach the per- cipient through sensory input, or as a kind of systematic noise conveyed by psi. Conditions that increase the likelihood of psi manifestation in the first place are generally likely to increase the likelihood of learning improved psi performance. While particular effects of the situation and experimenter bias on possible learning could be discussed at length (see Tart, 1977a), I only want to emphasize one thing here. Given that learning to use psi, as modeled above, is a subtle and difficult process, it is rather easy to interfere with it. Until our understanding of the learning process is much better, I think we should focus on optimizing the conditions which seem likely to produce learning, rather than being so concerned with their exact nature that we interfere with the process. PREDICTIONS By way of summary, I shall list the main predictions resulting from my application of learning theory to psi, specifically those predictions dealing with adequately motivated percipients who have a fair to high degree of psi talent at the start of immediate feedback training, and who train for periods of short to moderate length. 1. The vast majority of such percipients will show relatively stable psi performance, neither significantly inclining or declining. 2. Some percipients will show significant learning. 3. The degree of improvement of psi performance will be pos- itively correlated with the psi talent level of the percipient at the start of the training. These are the three predictions from the original presentation of my theory (Tart, 1966). Given the great importance of memory processes in building up adaptive response strategies, we can further predict that: 4. The early stages of training will be characterized by learning over short, homogeneous time periods such as the run, with decre- ments associated with interruptions of training such as breaks be- tween runs or sessions, interpolated activities that interfere with memory consolidation, and the like. 5. Sufficient training in the more talented percipients will allow learning to occur over longer time periods. 6. Procedures and factors which have generally been found to be psi-conducive will generally facilitate learning. Some other predictions are: 7. The following factors will be positively correlated with im- proved psi performance: (a) high (but not excessively high) motiva- tion, (b) high general learning ability, (c) absence of specific con- scious or (inferred) unconscious resistances to psi, (d) good ability to discriminate contents of the experiential field, (e) good ability to separate experience-as-perceived from experience-as-interpreted, (f) good memory skills, (g) ability to quiet one's mind; (h) non- attachment, ability to drop strategies that are not adaptive in spite of emotional investment in them, (i) low levels of maladaptive strategy boundness in the specific sense of not mechanically avoiding gues- sing what has just come up as the previous target, and (j) ability to ignore sensory distractions. In conclusion, it is well to remind ourselves of a prime problem of parapsychology: we want to advance our understanding of psi, but it usually works so unreliably and at such low levels that the study of its nature is very difficult. The signal is drowned in noise. The results obtained to date with immediate feedback training suggest that it is a valuable tool such that we can train reliable, high-level psi performance in some percipients and so open a door to profitable functional studies of the nature of psi. I hope others will join me in further elucidating the uses of this tool. REFERENCES HONORTON, C. State of awareness factors in psi activation. Journal of the American Society for Psychical Research, 1974, 68, 246- 256. KANTHAMANI, H., AND KELLY, E. F. Card experiments with a single subject. I. Single-card clairvoyance. Journal of Para- psychology, 1974, 38, l~26. (a) KANTHAMANI, H., AND KELLY, E. F. Awareness of success in an exceptional subject. Journal of Parapsychology, 1974, 38, 355- 382. (b) KENNEDY, J. E., AND TADDONlO, J. L. Experimenter effects in parapsychological research. Journal of Parapsychology, 1976, 40, 1-33. SCHMIDT, H. The psi quotient (PQ): An efficiency measure for psi tests. Journal of Parapsychology, 1970, 34, 210-214. SILVERMAN, I. The Human Subject in the Psychological Labora- tory. Elmsford, N.Y.: Pergamon Press, 1977. STANFORD, R. G. Response bias and correctness of ESP test re- sponses. Journal of Parapsychology, 1967, 31, 280-289. TART, C. T. Card guessing tests: Learning paradigm or extinction paradigm? Journal of the American Society for Psychical Re- search, 1966, 60, 46-55. TART, C. T. On the nature of altered states of consciousness, with special reference to parapsychological phenomena. In W. G. Roll, R. L. Morris, and J. D. Morris (Eds.), Research in Parapsychol- ogy 1973. Metuchen, N.J.: Scarecrow Press, 1974, Pp. 163-218. TART, C. T. The application of learning theory to ESP performance. - (Parapsychological Monographs No. 15.) New York: Para- psychology Foundation, 1975. (a) TART, C. T. States of Consciousness. New York: Dutton, 1975. (b) TART, C. T. Learning to Use Extrasensory Perception. Chicago: University of Chicago Press, 1976. TART, C. T. Toward humanistic experimentation in parapsychology: A reply to Dr. Stanford's review. Journal of the American Society for Psychical Research, 1977, 71, 81-101. (a) TART, C. T. Improving real-time ESP by suppressing the future: Trans-temporal inhibition. Paper delivered at a meeting of the Institute of Electrical and Electronic Engineers, New York, 1977. (b) TART, C. T. Psi: Scientific Studies of the Psychic Realm. New York: Dutton, 1977. (c) TART, C. T., PALMER, J. A., AND REDINGTON, D. J. Effects of immediate feedback on ESP performance: A second study. Sub- mitted for publication. TERRY, J. C., AND HONORTON, C. Psi information retrieval in the ganzfeld: Two confirmatory studies. Journal of the American Society for Psychical Research, 1976, 70, 207-217. WHITE, R. A. A comparison of old and new methods of response to targets in ESP experiments. Journal of the American Society for Psychical Research, 1964, 58, 21-56. WHITE, R. A. The influence of persons other than the experimenter on the subject's scores in psi experiments. Journal of the Ameri- can Society for Psychical Research, 1976, 70, 133-166. WHITE, R. A. The limits of experimenter influence on psi test results: Can any be set? Journal of the American Society for Psychical Research, 1976, 70, 333-369. Department of Psychology University of California, Davis Davis, California 95616