Conscious and Anomalous Nonconscious Emotional Processes: A Reversal of the Arrow of Time?

It should come as absolutely no surprise to anyone who has looked around my site that I am interested in “psi” type phenomenon.  I have not witnessed much beyond the mundane, myself.  But the “mundane” still would exceed what much of science has to say on the matter.  It excites me to see real research into the field, and it is why I am so impressed with Ingo Swann.

Ingo, more than any other person, has put thought to how “psi” works, what it is, and how to utilize it.  In short, it seems to be an amalgamation of subtle sensory inputs that provide deeper insight into the environment.  Organs such as the VNO (which picks up on phenomena type signals) and the magnetic organs in your hands and feet (that are sensitive to magnetic fields and part of what has caused the belief in “healing touch” type therapies) provide additional sensory inputs beyond the regular 5.  They are, by definition, “6th” and “7th” senses.

To make things weirder, there seems to be an ability to act independantly of both space and time.   The following is a lengthy, yet interesting, paper by a science duo at MIT.  It discusses precognition events, and provides some tantalizing evidence to its nature.  This ties into quantum physics, and concepts of time symmetry.  Enjoy:

Two previous experiments have been reported that tried to explore physiological indicators of “precognitive information” in which subjects respond prior to presented stimuli. In an elegant experiment in the early seventies, John Hartwell, then at Utrecht University, measured the Contingent Negative Variation (CNV) after a warning signal and before a random selected picture of a face was to be displayed (Hartwell 1978). The CNV is a brain potential that has been associated with anticipatory processes; more precisely the CNV is interpreted as a “readiness for response” preparation. The subjects in Hartwell’s studies were asked to respond with one of two buttons depending on the gender of the face on the picture. The warning stimulus was sometimes informative, that is, the subject could infer from the warning stimulus what the gender type of the face on the picture would be. In those trials a mean CNV was observed that clearly differed for the two stimuli categories. In the other case the warning stimulus was uninformative but it was hoped that the CNV still would indicate what type of picture was about to be shown. Such a finding would suggest that in some way or another the subject had nonconscious knowledge of the nearby future. 1

Nearly 20 years elapsed before the idea of precognitive information reflected in the physiology of subjects was picked up again by the second author of this article (Radin 1996). He used the physiological measures Skin Conductance, Heart Rate, and Plethysmography, which reflect behavior of our sympathetic and parasympathetic nervous system. Furthermore, in contrast to Hartwell, he used highly emotional pictures that were presented 5 seconds after the subjects had pressed the button for the next trial. In 3 independent studies Radin found significant differences in physiology, most notably in the skin conductance, preceding the exposure of calm versus extreme pictures. The precognitive response was termed “presponse.” Radin discussed a number of possible classical explanations for presponse but concluded that these do not apply.

However one potential “normal” explanation, namely the effect of anticipatory strategies, was not discussed at the time. Subjects who participate in this type of experiment while being aware that once every so often an extreme picture will be displayed may build up (generally incorrect) expectations about the probability that such an extreme picture will be shown in the forthcoming exposure. Indeed, owing to the “gambler’s fallacy,” their expectation may increase after each calm picture and decrease after an extreme. Superficially it appears that this could result in a mean anticipatory presponse that is smaller for calm stimuli than for extreme stimuli.

This possible explanation of the differences in presponse was later modelled through elaborate computer simulations by the first author and by an independent sceptical outsider. It turned out that the effect as described above only emerges when randomization is done without replacement, and therefore it could not explain Radin’s original results (see also discussion section). Thus the experimental results by Radin suggested a true, large and replicable “precognitive” psi effects with a remarkable signal to noise ratio

The first author of this chapter (DJB) was skeptical of these results and therefore decided to replicate the experiments using the same general procedure and the same picture material but completely different software and hardware and also a different randomization procedure. This would, if the effects could be replicated, make an explanation in terms of technical artefacts or inappropriate randomization less likely.

1. General procedure

A participant sits in a comfortable chair in a dimly lit room, the index and middle finger of the left hand connected to a skin conductance measurement device. In the instruction the experimenter emphasises that the subject should try to experience each trial as a complete new one. After the instruction and one or more demonstration trials, the experimenter leaves the room and the participant may start the first trial by pressing a key on the keyboard. After 7.05 seconds,2 a period that we call the foreperiod, a randomly chosen picture, either calm or highly emotional, is displayed for a specific exposure time (for randomization details see appendix 1). Before, during and after exposure the skin conductance is sampled by the computer with a sampling rate of 5 samples per second (see figure 31.1).

[Figure 31.1 here; figures not yet available]

Study 1:

This is a rather straightforward replication of the experiments reported by Radin (Radin 1996). However the following modifications were made. Rather than selecting pictures completely randomly from the total pool, pictures are organised in three sets with different ratios between calm and extreme pictures. The set is chosen randomly at the beginning of each experiment such that the ratio is even unknown to the experimenter. Then the pictures in the set are shuffled and presented in the shuffled order. Thus all pictures in the set are presented.

Furthermore, rather than having different exposure times between studies as reported in Radin’s original paper, we introduced this variable as a within subject variable. There are always two possible exposure times. Each of them is randomly selected with a probability of 0.5 .

Study 2:

This is a study with uninformed subjects. They are told that 10 control pictures will be presented in order to establish a baseline in the physiological behavior. According to this cover story the real experiment would begin after establishment of this baseline.

Unknown to these subjects there will be a single extreme picture presented at a random sequence number between 3 and 10. For about half of the subjects this extreme picture is erotic, for the other half the crucial picture is violent.

Study 3:

After study 2 the (same) subjects are now completely informed about the fact that occasionally there will be extreme pictures and thus the subject may have similar anticipation strategies as in study 1. Study 3 uses only one ratio (one extreme for every three calms) between extremes and calms, and moreover uses only 48 pictures from the original 120 pictures. These 48 pictures were selected on the basis of their stronger effects in study 1.

1.1 Stimulus material

The basic stimulus material is identical to the set of pictures that has been used by Radin (1996). It consists of 80 calm pictures and 40 extreme pictures of a violent and erotic nature. This basic set of pictures was slightly updated by the student experimenter in study 1. She adjusted the set for cultural differences, most notably replaced a few erotic pictures that would not be seen as very arousing in Europe with more extreme pictures. A snapshot review of the updated set and other information on this study is available through Internet ( For the studies 2 and 3 a subset of the 120 pictures of the basic set was used. This selection was based upon a qualitative evaluation of the response effect of each of the pictures in study 1. For the categories, violent and erotic, the most effective pictures were selected while for the calm category a random selection was made. In the category “violent” we included decorative body piercings, including genital piercings. The erotic pictures were of both homosexual and heterosexual nature. We did not make an effort to study the differential effects between these two subcategories.

1.2 Subjects

The subjects in all studies were recruited from the circle of friends or acquaintances of the experimenters. In the first study these were 16 health care professionals who followed a course in Therapeutic Touch. These subjects must be considered as very open minded. In the second and third study most subjects were students at the University of Amsterdam. Seven of the 32 considered themselves as skeptical toward the existence of paranormal phenomena.

1.3 Dependent variables

The dependent variable in both studies is the behavior of the skin conductance during the 7.5 seconds preceding the stimulus (the fore period). The between-sample correlations and the large number of data points presents us with a problem of how to collapse the skin conductance during the fore period into a single dependent value in order to prevent over-analysis.

We defined the dependent variable Pas the mean values of the samples between 4 seconds and 6 seconds after the subjects initiated the trial (which is indicated as the critical interval in figure 31.2) corrected with a baseline value obtained from the samples between 0.6 and 1.6 seconds after the subject started the trial.3

1.4 Independent variables

In the studies the following independent variables are used:

a. Type of stimulus (StimType); For each study this within subject variable has two values: calm and extreme. In the category extreme we discern two subcategories: violent and erotic. In study 2a the type of stimulus is a between-subject variable.

b. Exposure Time (Exp. Time); For each study this within subject variable has two levels.

c. Subject variables (Ss-X). The gender of the subjects may be an important variable because in normal research on the physiology of emotions gender-typical effects have been reported (Greenwald et al. 1989).

d. Ratio between calm and extremes (Study 1 only)

2 Hypothesis and explorations.

Because these three studies were basically done to validate the earlier results obtained by Radin, the major relevant hypothesis concerns the anomalous difference in anticipatory physiology before the exposure of calm and extreme pictures.

Furthermore we decided to explore the potential differences between different classes of emotional pictures and the effect of exposure time.

Finally we planned secondary analyses that could reveal normal sequential effects (study 1 only).

This was done because in the studies reported here a “randomization without replacement” was used.

  • We will explore the differential effects for calms and extremes matched for sequential position.
  • We will explore the effects on the presponse pattern distribution for different extreme: calm ratios.

3 Results of Study 1

3.1 Subjects

In study 1, 16 subjects were tested in the period between October 4 and November 14, 1996. Three were male, and 13 female. Their age ranged from 22 to 57.

3.2 Calm vs. Extreme effect

Each subject did 40 trials, so the total data set consists of 16 x 40 = 640 epochs of GSR. From these 640 there were 428 obtained with a calm stimulus and 212 with an extreme stimulus. These epochs are averaged for each stimulus category and a baseline, which is the mean value of the first sample (at the moment the subject presses the button to initiate the sequence), is subtracted.

In figure 31.2 the average response is given for the calm epochs and for the extreme epochs. No error bars are indicated because error bars presuppose normality of the data-distribution.

[Figure 31.2 here; figures not yet available]

The formal test consists of calculating the presponse P according to the definition given in the section on dependent variables and performing a Mann Whitney U test on the scores obtained preceding the calm and preceding the extreme stimuli. The resulting z-score is: 2.4 with a corresponding p-value of 0.016.

3.3 Violent vs. Erotic effect

In figure 31.3 we have plotted the mean presponse pattern for erotic pictures and for violent pictures (NB the mean calm presponse is subtracted from both means).

[Figure 31.3 here]

From the figure it appears as if the violent presponse comes early and is the largest. A comparison of the independent variable P for the two types of pictures yields a (Mann-Whitney U) z-value of 1.65 (p = 0.99). The results for the violent presponse alone are quite significant ( z = 2.94; p = 0.003).

[Figure 31.4 here]

3.4 Exposure time effect

In figure 31.4 the average difference in presponse between extreme and calm is given for long (3000 msec) and short (400 msec) exposures.

Although the differences are in the predicted direction, that is, the presponse effect is larger for short exposures (in contrast to the response effect) the differences when formally tested are near significant. (Mann Whitney U: z-value = 1.72, p = 0.085).

3.5 Secondary analysis

A number of secondary analysis were done in order to test for sequential explanations. No evidence was found for such an explanation in terms of sequential strategies. (See appendix 2.)

4 Results of study 2

4.1 Subjects

32 subjects, 16 males and 16 females, were tested in the period from February 17 to March 4, 1997. Their age ranged from 19 to 36.

4.2 Calm vs. Extreme effect

Each subject had only one extreme exposure, preceded by between 3 or 9 calm exposures. In figure 31.5 the difference between the mean physiological records of 32 extreme exposures and the mean of 184 calm exposures is given. It is clear that the effect is smaller than in study 1.

The formal test yielded a z-score for the difference between Calm and Extreme stimuli of 0.43. It should be noted that the power of study 2 was less than in study 1 (see also discussion).

[Figure 31.5 here; figures not yet available]

4.3 Violent vs. Erotic effect

Figure 31.6 shows the mean physiological record in cases of exposure of an erotic vs. cases with exposure of a violent picture. The over-all pattern is similar to the pattern found in study 1. The maximum value for violent exposures is earlier than for erotic exposures. There seems to be an interesting “symmetry” between presponse and response, especially for the erotic exposures.

[Figure 31.6 here]

The formal test yielded a z-score for the difference between Erotic and Violent pictures of 0.57 (n.s.).

4.4 Exposure Time effect

Figure 31.7 shows the calm vs. extreme effects split for the two exposure times. In contrast with the findings of study 1 it appears that the shorter exposure time does not have a presponse at all. It should be noted that in study 1 the shortest exposure time is 600 msec while in this study it is 400 msec.

[Figure 31.7 here]

The formal test yielded a z-score for the difference between the two exposure times of 0.67 (n.s.)

5 Results of study 3

5.1 Subjects

These were identical to study 2.

5.2 Calm vs. Extreme effects

Figure 31.8 shows the over-all results of study 3. The presponse is again clearly visible but although the power is even slightly higher than in study 1 the difference in Presponse between Calms and Extreme stimuli doesn’t reach statistical significance (z = 0.9; n.s.).

[Figure 31.8 here; figures not yet available]

5.3 Erotic vs. Violent effects

In figure 31.9 we may, with some fantasy, again discern an earlier presponse for violent than for erotic stimuli. The formal test however only deals with the magnitude of the effects during the critical period and hence do not show a significant effect (z = 0. 14; n.s.)

[Figure 31.9 here]

5.4 Exposure time effects

Figure 31.10 displays the mean GSR for extreme stimuli for short and long exposure times. The Mann Whitney U test yields a z-score of 0.55 (n.s.).

[Figure 31.10 here; figures not yet available]

6 Discussion

Table 31.1 reviews the three studies. The effect of a larger presponse for succeeding extreme pictures than for succeeding calm pictures is consistent through the three studies. We have calculated an effectsize measure (the difference in mean ranks for the two conditions that are compared normalised by the over-all mean rank in percent). This allows us to compare the studies quantitatively.

[Table 31.1 here; figures not yet available]

The Extreme vs. Calm effect is consistent throughout the three studies although the effectsize is only one third in studies 2 and 3. This effectsize reduction may be due to the different subject population. The combined result confirms the earlier findings of Radin.

The results of the comparison of violent and erotic stimuli in tabular form are confusing. The direction of the difference is not consistent across the studies but here, apart from the different subject population, the choice of the dependent variable may be questioned. In future research one should use a dependent variable that reflects the specific patterns (i.e., the early and the late phase of the presponse) better.

An interpretation of the exposure times findings is difficult because there were 3 different times used. It appears that an exposure time of 600 msec is better than 3000 msec but an exposure time of 400 msecs is inferior.

It should be noted that for 400 msec exposures the subjects do not always recognize the (emotional) contents of the pictures. So in that condition they may not always consciously experience an emotion. Radin (1997) has adequately treated a number of potential normal or classical explanations of the effect. The current replication with completely different hard- and software strengthens the conclusion that the results can not be explained by a technical artifact.

The major (and maybe only) source of normal explanations left after Radin’s original studies was the hypothesis that subjects developed anticipatory strategies that would result in artifactually different anticipatory physiology preceding calm or extreme pictures. As was explained in the introduction this seems at first sight a real possibility. The current results however hardly support this notion because the presponse effect seems not to be dependent on ratio between calm and extremes in a systematic way that should be expected if anticipation strategies based upon the Gambler’s Fallacy were employed.

There are three further arguments against an explanation in terms of normal anticipatory strategies. The first one of these is that we find suggestive internal effects that can not easily be explained by this type of strategies. For instance, the difference between erotic and violent stimuli. It would require anticipatory strategies that discriminate between the two types of extreme pictures to account for that effect.

The second argument is that an analysis that takes into account the sequential history gives basically identical results. Of course this analysis does not correct for possible strategies that are based upon doublets of extremes and the like.

The final argument is that computer simulations of anticipatory strategies, using ratios and total number of exposures that are also used in the current studies, do not show the expected main calm vs. extreme effects. It turns out that these simulations are extremely sensitive for the type of randomization. If we used a random selection with replacement (open deck) the effects were nil. If we used a random shuffling without replacement then the effects ranged between 0 and 10 percent. This finding came as a surprise because the reasoning as sketched in the introduction has such a direct appeal.

The following anticipatory strategies were tested:

a) increase anticipation by 1 unit after each calm and resets anticipation to 1 after each extreme.

b) double anticipation after each calm (with a maximum of 500) and reset the anticipation after each extreme to either half or 1.

The simulated effects in the open-deck situation were never larger than 2 percent while the experimental effects in studies 1 and 2 are generally larger than 10 percent.

However, these analyses are by no means exhaustive and there may be less plausible models that may result in larger differences.

The major point in favour of the psi hypothesis is however that there are no indications in the real data that support any of these sequential strategy models so far.

Is this effect an instance of macroscopic time symmetry?

The results indicate a precognitive response-subjects react up to several seconds before the stimulus appears. Having ruled out anticipatory strategies, the only remaining explanations suggest some backward time referral. Formally the laws of physics are time-symmetric. Practically this time symmetry is observed in classical mechanics but not in thermodynamics where Boltzmann’s second law forces the development of systems toward higher entropy. In a lucid book on time symmetry, Huw Price analyzes this problem and concludes that the standard “explanation” for assymetry based upon probabilistic arguments is incorrect (as Boltzmann himself also realized) (Price 1996). In an analysis of the asymmetry observed in radiation (EM theory) Price suggests that asymmetry here is caused by the spatial arrangements of radiation absorbers and emitters. Absorbers tend to be noncoherent while coherence is often the case for transmitters. According to Price this results in a destructive interference of any “advanced” waves (i.e., from absorber to emitter). Thus we never observe in nature back-action except potentially if we have a coherent absorbing system. Maybe our consciousness is such a system. Price shows also that when allowing for time-symmetry in quantum physics all puzzling paradoxes related to the measurement problem such as nonlocality disappear.

The results show that presponse occurs subconsciously but that (“subsequent”) conscious experience of the emotional figures are required. Price’s analysis of the problem of lost time-symmetry suggests a continuation of these types of experiments with experienced meditators in altered states. If the meditator succeeds in blocking out the picture of his awareness we may get a complete disappearance of the phenomena. Interestingly this fits with lore about the relationships between meditation and the occurrence of “psi”-phenomena. It is said that on the path toward complete control of one’s consciousness at some point psi-phenomena will appear. It is also said that one should not pay attention to these phenomena because that would only frustrate progress in meditation performance. Within this, admittedly very speculative, framework the expected point of symmetry on the time axis is NOT at stimulus onset but rather at the start of the conscious (emotional) experience, which may be around 500 msec later. Therefore the peak of the presponse is not expected around 3.5 seconds before the stimulus onset (where it would be if it was a mirror image of the response with symmetry point at stimulus onset) but rather about 2.5 seconds before stimulus onset, which fits well with the specification of the dependent variable in section 1.3.


Dagmar van der Neut was the experimenter in the first study. She was a continuous source of improvements and ideas. Also Rens Wezelman’s stimulating discussions and weird insights were and are instrumental in the success of the research program. The Parapsychological Institute offered hospitality for the carrying out of study 1.

Notes This article is an adapted version of a paper presented by the first author at the Parapsychological Convention (1997, Brighton UK).

1. The results of this experiment were, given the large efforts invested, disappointing. In hindsight it turns out that Hartwell’s sensitivity was too low (his design was such that the anomalous differential effect had to be larger than 30 percent of the normal differential effect in order to be detectable). Also he used stimuli that had only a very low emotional value.

2. Radin originally used a 5 seconds foreperiod.

3. In the original paper, presented at the PA 1997, the first author used a sightly different interval but new theoretical insights suggested to shift the critical intervall to one second earlier (see also discussion on time-symmetry). This does not affect the over-all conclusions.

4. GSR is used throughout the chapter because we measured the phasic response signal and hence slow changing tonic levels are filtered out. The GSR is therefore only given in relative arbitrary units.


Bechara, A., H. Damasio, D. Tranel, and A. R. Damasio. 1997. Deciding Advantageously before knowing the Advantageous Strategy. In Science 275:1293-1295.

Bierman, D. J., and D. I. Radin. 1997. Anomalous Anticipatory Response on Randomized Future Conditions. In Perceptual and Motor Skills 84:689-690.

Hartwell , J. 1978. CNV as an index of precognitive information. In European Journal of Parapsychology 2:83-103.

Greenwald, M. K., E. W. Cook, and P. J. Lang. 1989. Affective judgment and psychophysiological response: Dimensional covariation in the evaluation of pictorial stimuli. In Journal of Psychophysiology 3-1:51-64.

Merikle, P. M., S. Joordens, and J. A. Stolz. 1995. Measuring the relative magnitude of unconscious influences. In Consciousness and Cognition 4:422-439.

Murphy, T. S., and R. B. Zajonc. 1993. Affect, Cognition, and Awareness: Affective priming with Optimal and Suboptimal Stimulus Exposure. In Journal of Personality and Social Psychology 64 (5):723-839.

Price, H. 1996. Time’s arrow and Archimedes Point: new directions for the physics of time. Oxford University Press. ISBN: 0-19-510095-6

Radin, D. I. 1997. Unconscious perception of future emotions: An experiment in presentiment. In Journal of Scientific Exploration11 (2):163-180.

Proper randomization of the presentation order is a critical element in this type of experiments because the basic assumption is that the participant can by no normal means know what the following stimulus may be. Stimulus-arrays were shuffled using a pseudo random generator based upon the standard random function in the CodeWarrior C programming environment (CodeWarrior for Macintosh, version 8.). Sources of the software are available through Internet . It should be remarked that this randomization procedure is done at the start of the presentation of the first trial and therefore an interpretation of the results in terms of clairvoyance rather than precognition is allowed. (

A.1 Ratio effect

In order to evaluate the effect of different ratios between calm and extreme stimuli, an ANOVA was done using Stimulus category and Ratio as factors and the P-score as dependent variable. Given the nonnormality of the P-score distribution this turns out to be a slightly conservative approach. The results showed that the mean presponse for all stimuli, calms and extremes was heavily influenced by the ratio (F = 15.36, df = 2, p> 0.0001). This is because ratio is a between subject variable and subjects do differ greatly with respect to psychophysiological responsiveness. The interaction with stimulus category however was not significant suggesting that the ratio extremes:calms was of no influence on the calm vs. extreme presponse effect.

A.2 Matching for sequential position

In the previous analyses the means of the presponses were calculated independent of the sequential position of the specific stimulus. So the average of the calm presponses is composed of presponses of calm trials that were preceded by another calm trial but also of calm trials that were preceded by an extreme trial. This pooling of trials with different sequential position may result in artefacts as described in the introduction.

One solution to this problem is to compare trials only if they have an identical sequential history. The trials in study I were therefore broken down according to their sequential history. We then compared the last Calms and Extremes with the following histories:

lag1: Extreme-Calm vs. Extreme-Extreme

lag2: Extreme-Calm-Calm vs. Extreme-Calm-Extreme

lag3: Extreme-Calm-Calm-Calm vs. Extreme-Calm-Calm-Extreme

lag4: Extreme-Calm-Calm-Calm-Calm vs. Extreme-Calm-Calm-Calm-Extreme

The results for the different lags are graphically presented in figure 31.11. [Figure 31.11 here; figures not yet available] It can be seen that, although there are different wave forms for the different timelags, the over-all picture is the presponse for extremes is larger than for calms. Separate Mann-Whitney U tests yield the following z scores:

Lag z-score Ncalms 1Nextremes
1 0.30 146 77
2 1.74 96 46
3 2.86* 59 34
4 1.15 42 16

All further lags have also a positive z-score. A weighted sum of the 4 different analysis is given in figure 31.12 and shows basically an identical result as figure 2 [check number].

[Figure 31.12 here; figures not yet available]


This seems to support several, several previous experiments, and has very tight process and controls.  The one thing that Ingo said about his time doing “psi” experiments is that they are absolutely horrible and boring.  He began to help develop different types of experiments that were more engaging for him, and it brought success to the SRI team.  It seems that the MIT approach is lacking in any real mental excitement.  One wonders how it would change the results if the experiments themselves were more engaging?  By bringing the mind fully to bear….maybe it would help.  Then again, maybe not (as “psi” are very subtle senses that may be drowned out by an actively conscious mind).

Regardless, very interesting reading.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: