Human Electrodermal Response To Remote Human Monitoring: Classification And Analysis Of Response Characteristics - Statistical Data Included

PAUL STEVENS [1]

Abstract: This study reanalyzed datasets from 2 past studies and attempted to identify some characteristics of human electrodermal reactions to remote monitoring by another human, an aspect of direct mental interaction with living systems (DMILS) research. The objectives were (a) to see if an electrodermal DMILS response was similar to a sensory response and, if not, to see if there were any useful characteristics that could be used to identify the former and (b) to compare the electrodermal response seen in DMILS with that seen in reaction to a weak magnetic field, allowing exploration of potential mechanisms or physiological response systems that might produce the observed DMILS effects. No electrodermal activity (EDA) was observed that was obviously comparable to a sensory response, and there was no evidence of a consistent difference between activate and calm periods. Consistent between-participant differences were noted when comparing DMILS responsiveness to resting EDA. A consistent scale-invariant patt ern was found based on the variance of EDA, showing significant differences between any type of influence attempt and rest periods (p[less than]0.01 and p[less than]0.0002, 2-tailed). This pattern was also seen in the magnetic field exposure data, possibly indicating similarities between DMILS and magnetic response mechanisms.

Recent years have seen an increase in the use of physiological, as opposed to conscious, responses to ostensibly psi-mediated stimuli. Such research, often used in studies of direct mental interaction with living systems (DMILS), indicate that an individual's conscious response may not be a good measure of psi. Instead, a physiological reaction (often a measure of electrodermal activity; EDA) related to the stimuli is a more reliable indicator (e.g., Sah & Delanoy, 1994; Stevens, 1998). However, the DMILS research concentrating on physiological responses has tended to show an emphasis on using the responses purely as relative measures: The levels of physiological arousal in different conditions are compared with the mental intention (stare vs. not stare, or calm vs. arouse) of a remote person. There has been less research into the characteristics of response that is found -- information that could offer insights into possible mechanisms or artifacts. There are also many articles on the subject that contain t he implicit assumption that the electrodermal DMILS (EDA-DMILS) response might be akin to a sensory response -- for example, Braud, Shafer, and Andrews (1993) used psychological profiles for physiological responses to sensory stimulus to explain EDA-DMILS responses -- but there appears to be little work looking for those characteristics in the electrodermal data.

In the bioelectromagnetics field, it is noted (Bell, Marino, & Chesson, 1992; Conner & Lovely, 1988) that many organisms have a high sensitivity to certain electromagnetic field characteristics, usually in the low-frequency ranges that overlap biological activity (e.g., 0.5-30 Hz for global brain activity, 40 Hz thalamic-cortical loop, 100 Hz muscle activity). This has led some researchers (e.g., Popp, Chang, Gu, & Ho, 1994 [biophoton emission by organisms]; Ho, Ross, & Bolton, 1992 [electromagnetic synchronization between organisms]) to suggest that there may be an electromagnetic component to intercellular communication. If such communication does exist, then one might also expect a global response (based on the combined cellular response) in the human body in the presence of a suitable electromagnetic stimulus. Such a response would show up as a perturbation of physiological activity, and thus maybe also a behavioral change. An interesting question is therefore whether a DMILS response might be related to some form of interorganism electromagnetic communication.

Previous studies by Stevens (in press) showed a global response to an applied, weak magnetic field (MF). Participants exposed to a randomly occurring, oscillating MF (50 microTesla at 20 Hz) exhibited an average 2% decrease in level and 64% decrease in variance of EDA. This effect was not due to any conscious awareness of the fields or to external sensory cues. Affective perceptions were also perturbed during double-blind magnetic field (MF) exposure, with all images presented during field exposure being rated as more positive but less arousing than they were during control conditions. Compared to the typical EDA-DMILS response, this MF response was around 5 to 10 times stronger, but this was probably due to the much greater strength of the artificial field when compared with those that could be generated by the human body: At best, the MF of the latter is around 0.1 nanoTesla (i.e., more than 100,000 times weaker). Although this difference in magnitude is large, there is limited experimental support for fie lds comparable to the human biomagnetic field also having measurable effects on human physiology and behavior (e.g., Sandyk, 1992; Sandyk & Derpapas, 1994). So if the DMILS mechanism is associated with physiology-generated MFs, as other studies suggest (Sah & Delanoy, 1994; Schwartz, 1974; Stevens, 1998), then similar response characteristics might be expected in the recipients physiological. Essentially, EDA, and possibly also other response-measure DMILS experiments, may be seen as magnetic sensitivity experiments with a biological MF generator. By this, I mean that the physiological or behavioral responses seen in EDA-DMILS studies may relate to some form of electromagnetic interaction. This sort of explanation will be easier to apply to situations in which the 2 participants are relatively close together, as greater separation distances bring up currently unanswerable questions as to the possible range of such weak stimuli: Bioelectromagnetics research can offer no widely applicable mechanism for weak MF responses, so there is little basis for speculation either way as to their potential for long-distance interactions. Note also that I am not advocating a return to the early "mental radio" models that proposed encoded communication between individuals. Instead, I am suggesting that the presence of a biologically generated MF might have a significant effect on human physiology. An individual's interpretation of the effect will depend on a host of factors but not necessarily on the information content (if any) of that field.

Characterization and comparison of physiological responses to remote human monitoring and to an artificially generated weak MF might allow clearer definition of the EDA-DMILS effect, leading to better experimental design and insights into possible mechanisms for DMILS effects as a whole. It would also provide a greater understanding of the role that MFs play in normal physiological functioning, helping to determine the action of electronic equipment (e.g., video display units, fluorescent lighting, electric blankets) on humans. This would help clarify situations in which shielding would be beneficial to humans working with MF-generating equipment and may also lead to better or novel therapeutic uses of such fields.

ELECTRODERMAL RESPONSES TO SENSORY STIMULI

For an electrodermal response to a sensory stimulus, a rapid increase shortly (1-5 s) after the stimulus onset would typically be expected, followed by a slower decrease (Cacioppo & Tassinary, 1990). In an experimental setting, this phasic response would be superimposed on a downward trend (the tonic response) that indicated the percipient was getting more relaxed as the experiment progressed. Figure 1 shows an idealized form of this. If the EDA-DMILS response had a sensory component, then the presence of a similar, though probably weaker, electrodermal response might be expected.

Method

The data used were taken from existing physiological databases from two recent EDA-DMILS studies (Delanoy, Morris, Roe, & Brady, 1999; Watt, Ravenscroft, & McDermott, 1999) and a magnetic sensitivity study conducted by Stevens (in press). All of the data were recorded from self-selected volunteer participants. Skin conductance (microSiemens) was recorded using 78.5 [mm.sup.2] Ag-AgCl round electrodes and either an isotonic electrode paste or a water-based cream (the latter was used in the magnetic sensitivity study), placed on the second phalanx of the index and second fingers of the participant's nondominant hand and secured using a Velcro strip. Hardware data reduction was by means of the Physiodata monitoring system, model 1410 (J&J Engineering, Poulsbo, WA), which has a resolution of 0.24 microSiemens. This was interfaced to a high-speed serial port on a 100-MHz Pentium PC. Data were sequentially sampled at 1024 Hz, and time-averaged samples saved to disk at 16 Hz. In Stevens's study, test runs were cond ucted to ensure that there was no detectable direct pickup of the MFs by the physiology leads.

For the EDA-DMILS studies, the data consisted of arousal, calm, and rest (control) periods. Complete data were available for 43 participants in Watt et al.'s (1999) study and 80 participants in Delanoy et al.'s (1999) study. For the magnetic sensitivity data, the data consist of magnetic exposure versus null exposure (control) periods, and complete data were available for 29 participants. Both datasets used fully randomized, double-blind presentation of experimental versus control periods. All data analysis was conducted using Visual Numerics' PV-Wave numerical and graphical analysis programming environment.

One problem with measures of electrodermal activity is that different people will show different ranges of activity, making between-participant comparisons difficult. To avoid this, researchers often transform the raw skin-conductance values using some normalizing technique. In this study, the raw values were transformed to z scores, that is, expressed in units of the standard deviation (sigma, [sigma]) of each participant's skin-conductance values. This technique reportedly gives a useful and robust measure for subsequent analysis (Sersen, Clausen, & Lidsky, 1978). Thus, the nth raw data point for participant i would become:

[z.sub.ni] = ([x.sub.ni] - [x.sub.i])/[[sigma].sub.i]

Additionally, in the plotting of graphs, the initial value for each condition has been defined as the origin, as it is the relative change in skin-conductance values under each condition that is of interest.

RESULTS

An attempt was thus made to characterize the electrodermal response seen for each condition (activate, rest, and calm) in the two EDA-DMILS studies. Based on the expectations of sensory responses, each participant's data for the first 5 s of each experimental period were studied, but no obvious response was visually evident. A few participants did appear to occasionally exhibit an amplitude increase shortly after the presumed stimulus onset (i.e., the start of the influence period), but there was little consistency, making it more likely that these represented spontaneous, nonspecific responses. The possibility that there was a weak response that was lost in the noise was then investigated by combining the results of all participants in a study and plotting their average response profile. If there was a consistent but weak signal, this procedure should serve to amplify the consistent signal while the random noise would cancel out. The plots are shown in Figures 2 and 3.

Watt et al.'s (1999) study could be showing a DMILS response, although it is a very flattened one if so. There is little to distinguish between activate and calm profiles, although there is a clear distinction between influence and rest periods. Delanoy et al.'s (1999) study is less clear: The activate period looks more like the expected profile for the rest period, and both rest and calm periods show what could be a possible response to a DMILS stimulus.

As there appeared to be no classic sensorylike response in the expected period, this could mean that the EDA-DMILS effect occurs in a more subtle manner, showing an effect over the full influence period. To investigate this possibility, the averaged data for all of the participants from the entire duration of an experimental period was studied. Plots of these data are shown in Figures 4 and 5.

As can be seen, there is considerable variation in electrodermal activity over the full period. Watt et al.'s (1999) study profiles still appear to show a fairly clear distinction between influence and rest periods, but the activate and calm periods are again hard to distinguish. Delanoy et al's (1999) study profiles become even more confused although there is a suggestion of similar behavior for the influence profiles, but the direction with respect to rest is in the opposite direction to that seen in Watt et al.'s study.

Given the lack of any clear response in the initial 5s after stimulus onset, all of the following analyses refer to the whole epoch (20 or 30 s, depending on which study is being analyzed).

WATT ET AL.'S (1999) STUDY IN DETAIL

Table 1 shows the total number of individual's averaged skin-conductance responses in each direction (i.e., whether they were in the right or wrong direction with respect to the designated intention) for Watt et al.'s (1999) study. Activate periods seemed to show a better response rate, but calm periods were also more likely to show increased activity. Furthermore, within individual profiles, electrodermal activity for activate periods was higher than in calm periods in only 44% of cases. As the standard statistical measure used in EDA-DMILS is based on whether arouse periods show an increase and calm periods show a decrease in skin-conductance levels (i.e., the overall direction of the response), it is perhaps not surprising that past results have been sporadic.

This lack of consistency could indicate that there is no effect. However, this would be contrary to the overall evidence that there is a difference between activate-calm periods and rest periods. Looking again at the average profile for Watt et al.'s (1999) study (Figures 2 and 4), it is clear that the activate and calm periods show a consistent difference to the rest periods. An alternative approach would be to suggest that the DMILS stimulus was very weak so that the response was below the level of noise inherent in an individual's physiology However, this does not account for past findings which show that physiologically labile people exhibit a stronger effect (Braud, 1994). Such people should be far less likely to respond to a weak stimulus as they essentially have more noise in their systems. A more likely possibility is that DMILS does not work in a way analogous to a conventional sensory response. It may be a more basic form of interaction, or something possibly more akin to a direct influence.

The overall profile further suggests that the influence sessions (activate and calm) were more variable than the rest periods. This was tested by calculating the variance of each individual's skin-conductance values from each epoch, then combining them to give overall averages for that individual for each of the three conditions. Figure 6 shows the results.

A Wilcoxon signed-ranks test comparing the variance of each individual's skin-conductance values from each of the epochs (see Table 2) showed that all three conditions significantly varied from each other, with the greatest difference being between the calm-activate periods and the rest period.

DELANOY ET AL.'S (1999) STUDY IN DETAIL

Table 3 shows the direction of individual responses in Delanoy et al.'s (1999) study. This time, it was the calm periods that appeared to show the best response, with the activate periods more likely to show decreased activity. Within individual profiles, electrodermal activity for activate periods was higher than in calm periods in only 31% of cases (which accounts for this study failing to reach overall statistical significance based on a conventional EDA-DMILS analysis).

Figure 7 shows the comparison of the mean variance for each individual's responses for each epoch. The profile is very similar to that seen in Watt et al.'s (1999) study (see Figure 6) even though the scales are different. This suggests that there is a consistent pattern in the variance of the EDA-DMILS responses across individuals.

A Wilcoxon signed-ranks test (see Table 4) this time showed that both of the influence conditions significantly varied from the rest period but not from each other. As the original study was nonsignificant based on the activate-calm mean-level comparison, this could indicate that a more robust future measure would be to look at this variability of electrodermal activity in influence periods compared with rest periods, rather than at the mean level.

An interesting feature of Delanoy et al.'s study was that they also recorded electrodermal activity from the sender. This allows determination as to (a) whether the sender was trying to influence the receiver during the correct periods, (b) what kind of general strategies the sender was using, and (c) whether there is any correspondence between the sender's electrodermal activity and the receiver's. The averaged profiles for the initial 5 s and the full duration are shown in Figures 8 and 9, respectively.

Interestingly, the senders' electrodermal activity is very similar irrespective of whether they are attempting to activate or to calm the receiver.' This implies that they were using an active strategy (i.e., getting worked up while trying to achieve their aims) rather than attempting to simulate the desired state in themselves as is often assumed.

INDIVIDUAL PROFILES

Looking at the DMILS datasets as a whole, it was noticed post hoc that the individual profiles appeared to fall into two broad categories: low and high responders. The average electrodermal activity profiles for individuals in each of the groups showed similarities. Low responders had very flat profiles, with little change between rest, activate, and calm periods. High responders were more variable, showing strong (though not necessarily consistent) differences between the conditions. This appeared to be true whether the data considered were for the initial 5 s (during which one would expect a sensory response to occur) or over the full duration of each DMILS condition. Some typical examples of low and high responders in the two studies are shown in Figure 10.

All of the participants were then numerically classified as responders versus null responders, depending on whether they exhibited any kind of response greater than the (arbitrary) value of 0.2 sigmas at any point after the start of the influence period. A Kolmogorov-Smirnow two-sample test was used to compare the distributions of their resting physiological activity to see if they were significantly different (i.e., whether the baseline physiological activity of responders was different from null responders). Results are given in Table 5.

It appears that there was some a priori difference between those who showed a response and those who did not, the null responders having a lower variability to their resting physiological activity (i.e., they are less labile). This could be a trivial finding if all the physiological fluctuations were nonspecific and not due to any kind of DMILS effect; in effect, it shows that people who exhibit fluctuating physiology do so consistently. However, given that both of the original studies showed a significant difference in physiological activity between conditions (albeit in the wrong direction for Delanoy et al.'s, 1999, study), it seems worthwhile to suggest that physiological lability has a role to play in DMILS effects. For example, this could indicate that future studies would benefit from preselecting participants who had higher lability because these participants, if they showed a DMILS effect, would show a stronger response.

Note that there are some problems with this analysis in that the rest periods were interspersed with the influence periods; as such, they are not independent measures. A better measure would be to collect pre-experiment baseline physiological activity, using this to screen participants for further involvement. Note also that this analysis was based on visual inspection of a response in the averaged profiles. The null responders may also include participants who showed weaker responses, so more research is needed to better define a labile physiology.

COMPARISON WITH RESPONSES TO MAGNETIC FIELDS

For the data from the magnetic sensitivity study, a sensory-like response was again looked for in the 5 s of each experimental period (note that the MF exposure periods, unlike the two DMILS studies, were only 5 s in length in total). As with the DMILS, no clear response of this type was found. Figure 11 shows the overall averaged response profile. There appears to be a slight depressing of the electrodermal activity during field exposure periods, which corresponds to the statistical analysis performed in the magnetic sensitivity study (Stevens, in press), which indicated the mean level of electrodermal activity was lower during field exposure.

Figure 12 shows the result of calculating the mean variance for each individual's responses for MF and control exposure periods. Note that the profile is very similar to that seen in the two DMILS studies (equating MF with activate and control with rest).

A Wilcoxon signed-ranks test comparing the differences in variance (see Table 6) showed that the two conditions did not significantly vary from each other, although they were in the prespecified direction (note that in the MF study a unidirectional response was expected based on earlier work; see Stevens (in press) for details). The nonsignificance could have been due to the lower number of participants used in the MF study.

DISCUSSION

The data from the first 5 s of each experimental period showed no clear response after the presumed onset of the DMILS stimulus. Moreover, the combined results showed little to distinguish between conditions. There were clear differences within each of the two studies, but this was not consistent between studies. The data from the entire duration of an experimental period also showed considerable variation, but there was slightly more consistency between studies. Watt et al.'s (1999) study profiles still showed a distinction between influence and rest periods, and Delanoy et al.'s (1999) study profiles showed a suggestion of similar behavior, although in the opposite direction. Overall, it does not appear that DMILS works in a way analogous to a response to a sensory stimulus. Instead, it may represent a more basic form of interaction, possibly more akin to a direct influence (i.e., the original idea of biological psychokinesis). Another possibility is that the DMILS stimulus, if such exists, takes longer to be detected than the 5-s window used in the initial analysis allows. This does appear to be the case in certain instances, as some individuals show a maximal difference between calm and arouse conditions in the 5-10-s window. However, even if this were the case, such individuals' responses rarely show the expected return to baseline that should occur after they have habituated to the stimulus, so again the response seems dissimilar to a classical sensory one.

Furthermore, some individuals appear to respond to DMILS in ways that are self-consistent but which do not necessarily correspond to the intended direction. The ones who are responsive do, however, seem to react to any type of influence attempt in a different way than is seen during rest periods. As the standard statistical measure used in EDA-DMILS is based on the calm-activate difference, it might perhaps be better to use an influence versus rest analysis. This would not tell us anything about how the intent to affect the target system in a specific direction might affect that system, but it might increase the reliability of the basic DMILS effect, allowing better theorizing as to possible mechanisms. Once more is known about potential mechanisms, then the more complex area of directional intention could be studied with greater confidence. Such a conclusion is also borne out by other DMILS studies that use a variety of response measures. For example, Braud et al. (1993) reported on findings using a simple staring versus no-staring protocol (comparable to an influence vs. rest protocol). Theyfound that a variety ofstudies showed significant results but that even with this simple protocol, the direction of the effect was not always consistent, apparently altering in response to the participants' attitudes to being stared at. This is again seen in Schlitz and Braud (1997), in which a summary of 15 electrodermal-response studies showed that four of the direction-specific studies exhibited reversed effects. A further study by Radin and Taylor (1993) shows a plot (see their Figure 2) of example data wherein the difference between either type of influence period and rest periods is much greater than the difference seen when comparing influence periods to each other. Although Radin achieved an overall significant finding, one wonders whether a greater and more consistent effect might have been revealed had an influence versus rest protocol been used.

One new finding was that the skin-conductance responses recorded during influence periods were significantly more variable that during the rest periods in both studies, the calm period showing the highest variability, then the activate period, then a large drop for the rest periods. The consistency across studies suggests that there is a consistent pattern to EDA-DMILS responses across individuals based on the variance of their responses. Once again, this indicates that future studies might benefit from using an influence versus rest analysis, but also that taking the variability of the responses into account might result in a more robust measure.

A potential problem with the activate versus calm protocol was demonstrated by the senders' physiological activity taken in Delanoy et al.'s (1999) study in which it was seen that the senders' electrodermal activity is similar irrespective of whether they are attempting to activate or to calm the receiver. This implies that the senders were using an active strategy (i.e., getting worked up while trying to achieve their aims) rather than attempting to simulate the desired state in themselves, contrary to the expectations of researchers in the field (e.g., Braud, 1994). If DMILS is a purely mental undertaking, then this may not be a problem. However, if there is any physical signal involved in the process (i.e., the sender and receiver are involved in some sort of energetic exchange), then this could indicate that there will be a difficulty in distinguishing between calm and activate "signals." Presumably any signal would relate to energetic processes in the body, of which physiological activity might be a goo d indicator on a gross level (physiological arousal tends to correlate with mental arousal). If the sender is experiencing near identical levels of physiological arousal in both influence conditions, then one might expect any DMILS signals to be more similar, at least compared with the lack of signal during the rest periods. It seems likely that an untrained receiver would find it easier to differentiate between the signal/no-signal periods than between two high-arousal signals.

To compare the EDA-DMILS responses to a known stimulus, the data from Stevens's (in press) magnetic sensitivity study was studied. As with the DMILS data, no clear sensorylike response was found, but there was an overall slight depressing of the electrodermal activity during field exposure periods. Looking at the mean variance for each individual's responses, the profile was similar to that seen in the DMILS studies (equating MF exposure periods with activate, and control periods with rest), although the distribution of responses was not as widely separated as with the DMILS conditions and the difference was not statistically significant (possibly due to the lower number of participants used in the MF study). Although not identical, there were a sufficient number of commonalities between the DMILS and MF study responses to encourage further explorations of possible electromagnetic-related mechanisms.

It also appeared that there could be some a priori difference in resting physiological activity between those who showed a physiological response after the onset of the DMILS stimulus and those who did not. It was thus suggested that future studies could benefit from preselecting participants who had higher lability of their resting physiology, although more research is needed to better define the concept of the ideally labile system: Although some degree of lability might be a good thing, too much runs the risk of losing any effects in the inherent noise of the system.

RECOMMENDATIONS FOR FUTURE STUDIES

* Replace the activate versus calm protocol and instead compare simpler influence versus rest periods.

* Include a measure of variance of electrodermal responses in the analysis.

* Pay more attention to the sender's physiological reactions. That is, measure the way in which the sender reacts when they are trying to have an effect and only use strategies which actually show a different physiological response in them first.

* Pre-select receivers based on high physiological lability.

* Continue research into possible electromagnetic factors.

(1.) I would like to express my gratitude for the support of a Perrott-Warrick Research Grant for this research. Thanks are also due to Caroline Watt, John Ravenscroft, Zachary McDermott, Robert Morris, Deborah Delanoy, Claire Brady and Alison Roe for allowing the use of their experimental data, and to John Palmer and the anonymous reviewers for their helpful comments.

(2.) The effect size measure, Cohen's d, was calculated from the observed Wilcoxon rank sum by transforming the smallest rank sum, W, to the approximate normal deviate, z, using the formula z = [0.25 N (N+ 1) - W - 0.5] / [[(N (N+ 1) (2N+ 1) / 24].sup.0.5], given in Snedecor & Cochran (1980). The z value was then transformed to Cohen's d using the formula d = [2 z [N.sup.-0.5] / [[N-1].sup.0.5], given in Rosenthal & Rosnow (1991).

REFERENCES

BELL, G. B., MARINO, A. A., & CHESSON, A. L. (1992). Alterations in brain electrical activity caused by magnetic fields: Detecting the detection process. Electroencephalography and Clinical Neurophysiology, 83, 389-397.

BRAUD, W. G. (1994). Reaching for consciousness: Expansions and complements. Journal of the American Society for Psychical Research, 88, 185-206.

BRAUD, W., SHAFER, D., & ANDREWS, S. (1993). Further studies of autonomic detection of remote staring. Journal of Parapsychology, 57, 391-409.

CACIOPPO, J. T., & TASSINARY, L. G. (1990). Principles of psychophysiology. Cambridge University Press.

CONNER, M., & LOVELY, R. (EDS.) (1988). Electromagnetic fields and neuro-behavioural function. New York: Alan R. Liss.

DELANOY, D. L., MORRIS, R. M., ROE, A., & BRADY, C. (1999). An EDA DMILS study exploring agent-receiver pairing. Proceedings of Presented Papers: The Parapsychological Association 42nd Annual Convention, 68-82.

HO, M.-W., Ross, S., & BOLTON, H. (1992). Electrodynamic activities and their role in the organisation of body pattern. Journal of Scientific Exploration, 6, 59-77.

POPP, F.-A., CHANG, J.-J., GU, Q., & HO, M.-W. (1994). Nonsubstantial biocommunication in terms of Dicke's theory. In M.-W. HO, F.-A. Popp, & U. Warnke (Eds.), Bioelectrodynamics and biocommunication (pp. 293-318). Singapore: World Scientific.

RADIN, D. I., & TAYLOR, R. K. (1993). Remote mental influence of human electrodermal activity: A preliminary replication. Proceedings of Presented Papers: The Parapsychological Association 36th Annual Convention, 12-23.

ROSENTHAL, R., & ROSENOW, R. (1991). Essentials of behavioral research: Methods and data analysis. Singapore: McGraw-Hill.

SAH, S., & DELANOY, D. (1994). Cognitive and physiological psi responses to remote positive and neutral emotional states. Proceedings of Presented Papers: The Parapsychological Association 37th Annual Convention, 128-138.

SANDYK, R. (1992). The influence of the pineal gland on migraine and cluster headaches and effects of treatment with picoTesla magnetic fields. International Journal of Neuroscience, 67, 145-171.

SANDYK, R., & DERPAPAS, K. (1994). The effects of picoTesla range magnetic fields on visuospatial functions in multiple sclerosis. International Journal of Neuroscience, 78, 103-109.

SCHWARTZ, G. E. (1974). Biofeedback and pattern self-regulation in biological perspective: A critical analysis of extra-sensitive perception [Abstract]. In J. D. Morris, W. G. Roll & R. L. Morris (Eds.), Research in parapsychology 1973 (pp. 199-229). London: Scarecrow Press.

SCHLITZ, M., & BRAUD, W. (1997). Distant intentionality and healing: Assessing the evidence. Alternative Therapies, 3(6), 62-72.

SERSEN, E. A., CLAUSEN, J., & LIDSKY, A. (1978). Autonomic specificity and stereotypy revisited. Psychophysiology, 15, 60-67.

SNEDECOR, G. W., & COCHRAN, W. G. (1980). Statistical methods. Ames: Iowa State University Press.

STEVENS, P. (1998). Techno-dowsing: Developing a physiological response system to improve psi training. Journal of Scientific Exploration, 12, 551-568.

STEVENS, P. (in press). Effects of 5-second exposure to a 50 mT, 20 Hz magnetic field on skin conductance and ratings of affect and arousal. Bioelectromagnetics.

WATT, C., RAVENSCROFT, J., & MCDERMOTT, Z. (1999). Exploring the limits of direct mental influence. Journal of Scientific Exploration, 13, 515-536.

                 PERCENTAGE OF SKIN-CONDUCTANCE RESPONSES
         IN INTENDED DIRECTION OVER WHOLE EPOCH BY DMILS CONDITION
                            (WATT ET AL. STUDY)
Condition Percentage of responses
                 in right         in wrong  no consistent
                 direction        direction   response
Activate            37               28          35
Calm                28               44          28
                   WILCOXON SIGNED-RANKS TEST COMPARING
                 VARIANCE OF SKIN-CONDUCTANCE VALUES OVER
                    WHOLE EPOCH IN EACH DMILS CONDITION
                            (WATT ET AL. STUDY)
                   Effect Size
                 (Cohen's d) [2] N  p value (2-tailed)
Activate vs Calm      0.18       43  [less than].0001
Activate vs Rest      0.22       43  [less than].000001
Calm vs Rest          0.2        43  [less than].00001
                 PERCENTAGE OF SKIN-CONDUCTANCE VALUES IN
                  INTENDED DIRECTION OVER WHOLE EPOCH BY
                  DMILS CONDITION (DELANOY ET AL. STUDY)
Condition Percentage of responses:
                  in right         in wrong  no consistent
                 direction         direction   response
Activate             34               58           8
Calm                 51               38          11
                   WILCOXON SIGNED-RANKS TEST COMPARING
                 VARIANCE OF SKIN-CONDUCTANCE VALUES OVER
                    WHOLE EPOCH IN EACH DMILS CONDITION
                          (DELANOY ET AL. STUDY)
                  Effect Size        p value
                 (Cohen's d) [2] N  (2-tailed)
Activate vs Calm     0.03        80   .19
Activate vs Rest     0.10        80   .0002
Calm vs Rest         0.13        80   .0002
                  KOLMOGOROV-SMIRNOV COMPARISON OF REST-
                     PERIOD PHYSIOLOGICAL ACTIVITY OF
             INFLUENCE-PERIOD RESPONDERS AND NULL- RESPONDERS
                     Mean variance of                    p value
                     skin-conductance N  [D.sub.m,n]   (2-tailed)
Watt et al. study
 Null Responders          0.0006      29   0.7190    [less than].001
 Responders               0.0026      14
Delanoy et al. Study
 Null Responders          0.0023      54   0.41      [less than].01
 Responders               0.0091      26
                   WILCOXON SIGNED-RANKS TEST COMPARING
                  VARIANCE OF SKIN-CONDUCTANCE VALUES BY
                 EXPOSURE CONDITION (MF SENSITIVITY STUDY
               Effect size
               (Cohen's d) N  p value (1-tailed)
EMF vs Control    0.11     29        .09

COPYRIGHT 2000 Parapsychology Press
COPYRIGHT 2001 Gale Group