Dara Deshe and Eugene Gokhvat

Dara Deshe and Eugene Gokhvat PSY/ORF 322 Spring 2005 Final Research Paper

Group Intention Effects on ShortREG

The Princeton Engineering Anomalies Research (PEAR) Laboratory has been conducting anomalies research for the past 26 years, finding empirical evidence that the human conscious may have an effect on machine outputs. The concept itself is astounding, however the idea that the mind can control a machine is being utilized and implemented in many areas of science today. For example, an article from BBC News

Online (http://news.bbc.co.uk/1/hi/health/4275245.stm ) entitled “Brain-controlled ‘robo- arm’ hope” stated that research and development being done at the University of

Pittsburgh is making a robotic arm to be controlled entirely by thought. Thus far human tests have shown “corresponding patterns of brain signals linked to limb movements.” Dr.

Andrew Schwartz, University of Pittsburgh, and his team are currently testing the effects on monkeys by inserting very small probes into the brains of the monkeys. These probes then “interpret signals from individual nerve cells in the motor cortex.”

While the article says the doctors believe they are about four years away from having a usable mechanical arm that would also be able to have functioning elements like individual fingers, the idea that the mind can control a machine is being tested in a much more mystical way in PEAR. The project was instigated by the work of physicist Helmut

Schmidt, who had done experiments “suggesting that devices involving random physical processes could be influenced solely by the subjective intentions of their human operators” (Dunne, 3). Through the use of various types of Random Event Generators

(REG), PEAR has acquired an enormous database of experimental data. Results from their “benchmark” database which is comprised of nearly 750,000 trials per intention has

“displayed positive correlations with operator intention that were statistically significant at a level of chance probability less than 10^-4, even though the observed effect sizes were quite small, of the order of 10^-4 bits deviation per bit processed” (Dunne, 22). By

“intention” here, we mean whether the operator intended to shift the mean to a higher or lower value, or to a stronger baseline value of 100 (with less variance).

Human intention is the primary variable being studied in these experiments. The intention of our research is to look at the results of group intention using data from past

PSY/ORF 322 classes, as well as data from visiting classes from Hillsboro Elementary

School and Tower Hill School. The following table indicates how our data is compiled and the approximate details of each group:

Group Data Set Series Dates Conducted Operators Age Range PSY/ORF 322 HIP.712 6 11/11 – 11/19/1998 3 – 10 19-22 PSY/ORF 322 HIP.921 12 03/28 – 04/14/2005 N/A 19-22 Hillsboro School HILLSBOR.703 7 02/24 – 03/11/1999 ~12 4th Grade Hillsboro School HILLSBOR.704 10 03/21 – 05/26/2000 ~12 4th Grade* Hillsboro School HILLSBOR.716 8 03/23 – 05/31/2001 ~12 4th Grade Tower Hill School TWRHLL.720 1 35031?? ~15 High School [Calibration Data] CALIB.991 33 (older) N/A N/A [Calibration Data] CALIB.995 6 (recent) N/A N/A

*Note: Because we only know the male/female ratio of the current PSY/ORF classes, that data will be our only basis of conclusion on male/female intentions. Also, in HILLSBOR.704 the 10th series is from a class of second grade students.

In Margin’s of Reality, Professor Jahn says “… deliberate investigations of group interactions with physical systems and processes, beginning with multiple operator experiments…or others testing various implications of the model for group behavior, could be instructive,”(328). For our purpose, this excerpt is very encouraging. The only other notable information we found regarding group intention was also in Margins, in an analogy of group interactions to those interactions of organic molecules, chains or rings.

“Here the anomalous group consciousness effects would correspond to the various modes of information and energy transfer and storage characteristic of such integral macromolecular structures, that are also inexplicable by any simple combination of the individual atomic behaviors ,”(256).

While many experiments have been done with single operators and co-operators of various relation to each other, we found no evidence of group intention analysis, only some available data which is what we are using for this project.

Works Cited:

Dunne, B.J. & R.G. Jahn. “Consciousness, Information, and Living Systems.” PEAR.

Jahn, R.G. & B.J. Dunne. Margins of Reality. 1987.

Roberts, Michelle. “Brain-controlled 'robo-arm' hope.” bbc.co.uk. http://news.bbc.co.uk/1/hi/health/4275245.stm Data

Below you will find the data found using the ShortReg Analysis Panel in the PEAR laboratory computers. Table 2.0 and Table 3.0 list the cumulative means and standard deviations of each data set. In these tables, the Hillsboro School group HILLSBOR.704 is divided into two subgroups in order to distinguish between series 1-9 which consists of fourth grader’s data, and series 10 which consists of second grader’s data. Table 4.0 lists each group’s Chi Squared data for high, low, and baseline intentions. *Note: for all tables the CALIB.991 and CALIB.995 data are listed under “Baseline” since there is no intention for the calibration data. Below these tables are the accompanying graphs and additional graphs containing a combination of all the data from each of the respective groups: Hillsboro School, PSY/ORF classes, and Calibration data.

Table 2.0: MEANS Group Data Set Series High Baseline Low PSY/ORF 322 HIP.712 6 99.735 99.88917 100.135 PSY/ORF 322 HIP.921 12 100.01 100.03208 100.26333 Hillsboro School HILLSBOR.703 7 99.72571 99.67214 99.80857 Hillsboro School HILLSBOR.704 series 1-9 100.087 99.911 99.9755 Hillsboro School HILLSBOR.704 series 10 99.235 100.485 100.185 Hillsboro School HILLSBOR.716 8 99.82125 99.9625 99.98 Tower Hill School TWRHLL.720 1 98.8 99.67 100.42 [Calibration Data] CALIB.991 33 NA 99.9913 NA [Calibration Data] CALIB.995 6 NA 99.95461 NA Total Hillsboro 703, 704, 716 24 99.235 100.485 100.185 Total PSY/ORF 712, 921 18 99.9345 100.00425 100.15 Total Calibration 991, 995 39 NA 99.98712 NA

Table 3.0: STANDARD DEVIATIONS Group Data Set Series High Baseline Low PSY/ORF 322 HIP.712 6 6.93456 6.96376 6.93447 PSY/ORF 322 HIP.921 12 7.0697 7.18541 7.06864 Hillsboro School HILLSBOR.703 7 7.18392 7.14341 7.21669 Hillsboro School HILLSBOR.704 series 1-9 7.07032 6.9466 7.27677 Hillsboro School HILLSBOR.704 series 10 7.46043 6.78083 7.72328 Hillsboro School HILLSBOR.716 8 7.08224 7.03914 7.06741 Tower Hill School TWRHLL.720 1 7.09094 7.16784 6.90449 [Calibration Data] CALIB.991 33 NA 7.07525 NA [Calibration Data] CALIB.995 6 NA 7.0724 NA Total Hillsboro 703, 704, 716 24 7.09073 7.0411 7.1707 Total PSY/ORF 712, 921 18 7.07444 7.12646 7.06863 Total Calibration 991, 995 39 NA 7.07492 NA

Table 4.0: Chi Squares Group Data Set Series High Baseline Low PSY/ORF 322 HIP.712 6 1.69 0.29 0.44 PSY/ORF 322 HIP.921 12 0 0.5 3.33 Hillsboro School HILLSBOR.703 7 2.11 3.01 1.03 Hillsboro School HILLSBOR.704 series 1-9 0.3 0.32 0.02 Hillsboro School HILLSBOR.704 series 10 2.34 0.94 0.14 Hillsboro School HILLSBOR.716 8 1.02 0.04 0.01 Tower Hill School TWRHLL.720 1 5.76 0.44 0.71 [Calibration Data] CALIB.991 33 NA 0.42 NA [Calibration Data] CALIB.995 6 NA 1.48 NA Total Hillsboro 703, 704, 716 24 4.32 3.9 1.12 Total PSY/ORF 712, 921 18 1.03 0.09 3.39 Total Calibration 991, 995 39 NA 1.91 NA

Graphical results in order of age group: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture.

QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture.

Z-Score for Aggregate Data:

Total Group Effect Statistic High Baseline Low N Trials 8800 8800 8800 Mean 99.88295 99.90693 100.06011 Z of Mean -1.55278 -1.23469 0.7975 Std. Deviation 7.07407 7.06697 7.11819 Z of SD 0.05629 -0.0768 0.88408

Z-Score = (1.55278+ 0.7975)/(2^.5) = 1.6619

Specific Class Data: Means w/ Respecitive Standard Deviation PSY/ORF 322 S'05 High Baseline Low Grou # # Total St. St. St. p Boys Girls # Mean Dev. Mean Dev. Mean Dev. 99.3 7.0632 99.39 7.1589 99.06 6.772 1 3 1 4 99.36 7.7035 100.48 7.8759 100.39 6.416 100.96 7.233 98.93 7.6624 99.62 7.88 2 3 4 7 100.49 6.8586 99.18 6.3634 99.32 7.418 100.04 7.0952 99.75 7.216 100.29 7.249 3 2 3 5 99.25 7.3214 100.21 7.0629 100.72 6.781 99.45 6.8555 100.42 6.6533 100.47 7.11 4 4 1 5 100.34 7.3803 101.03 6.545 100.49 7.334 100.9 6.5636 99.67 6.8579 100.33 2.638 5 2 3 5 100.95 6.8111 99.67 6.921 99.96 7.05 99.29 6.7864 99.03 7.6613 101.18 6.419 6 3 1 4 100.46 7.5417 101.44 6.6656 100.93 6.939 100.25 7.0873 100.47 7.5951 100.16 7.225 7 2 1 3 98.5 7.6719 101.52 6.8039 100.19 6.413 99.46 7.918 101.69 7.5018 100.31 7.031 8 5 3 8 100.05 6.6308 100.63 7.4815 99.27 7.039 100.06 7.068 99.21 6.6139 99.93 6.977 9 5 1 6 101.29 6.9679 100.36 6.9971 100.33 6.8 100.15 7.0787 99.51 7.4759 101.02 7.829 10 7 1 8 100.41 7.0095 99.85 7.234 101.56 6.936 99.05 6.649 100.69 7.6419 99.51 6.891 11 4 6 10 100.23 6.8222 100.5 7.6337 100.24 7.135 100.76 6.9865 99.2 6.7659 101.13 7.089 12 4 1 5 99.24 6.3789 98.94 7.3686 99.91 7.144 Chi-Square Summary Data:

Chi Square Analysis of PSY/ORF 322 Spring 2005 Data Intention Chi-Square DF High 24.0944 24 Baseline 32.5726 24 Low 21.87 24 TOTAL 78.537 72

*For further details on how we arrived at these figures please see the attached excel document on the last page. Analysis and Conclusions

Analysis of Calibration Data:

CALIB.991: Although there are erratic outliers which are exaggerated by the scale of the graphs, the means and standard deviations of the calibration data are nearly 100 and 7.0717, which has been the benchmark for all consequent PEAR experiments CALIB.995: This data seems to fit the Gaussian Distribution better than CALIB.991. The outlier can be explained by scaling. The mean and standard deviation, however, is consistent with the older calibration data. Combined: From this we can conclude that the aggregate calibration data indeed fits the expected baseline mean and standard deviations, thus ensuring that no responsibility of distinctive results could be attributed to a faulty machine of any sort.

Analysis of Groups according to Age Group:

HILLSBOR.704 Series 10: There is a negative mean shift on the high intention, even beyond the 95% confidence interval. The baseline does have a higher mean shift than the low intention line, however because we only have one series of data for this group, our results are in no way statistically significant.

HILLSBOR.703: At the peak of this graph, the lowest of highs is much lower than that of the other intentions. In the second half of the data there are extremely opposite results from the given intentions.

HILLSBOR.704 Series 1-9: The data for the high intention has a higher mean shift and the low intentions have a negative mean shift, however the baseline data has an even more negative mean shift than the low intention which is counter-to its given intention.

HILLSBOR.716: Initially, the high and low follow their respective intentions. The initial high intention, however, degrades with time causing the high intention to have the most negative mean shift.

Combined: The overall baseline intentions are have negative mean shifts, and while the high is relatively higher than the low intentions, all means in this case are slightly below the baseline value.

TWRHLL.720: This data appears to go entirely against their intentions. While this does fall within a slight pattern we are finding, because we have only one series (of 6 runs) this data is not statistically significant enough to draw any usable conclusion.

HIP.712: The mean shifts are opposite the intentions, however , the means are below baseline. HIP.921: In the 2005 class data, we found the same opposite relationship relative to the given intentions. In this case, unlike the 1998 data, the means are located above the baseline, showing an overall positive mean shift.

Combined: The fact that both classes together show an even more reverse mean shift than their intentions magnifies the reverse effect we have been continuously observing.

Total Group Effect: The overall picture depicted by the combination statistics and graph of all the of groups indicates that there is a statistically significant mean shift in the direction opposite to intention. The end result is a low mean shift for the high intention and a high mean shift for the low intention. According to the z-score of the data, which is 1.66%, this means as a whole the data is significantly backward to the previous work PEAR has done.

PSY/ORF Spring 2005 Gender Analysis:

In previous REG experiments, the general finding is that most of the females’ low intention results tend high. To test this idea, we looked at groups with more females than males. While there were only four groups dominated by females, we still could not draw such a conclusion. Perhaps this is because no group was truly female-dominated (since the ratio was often 3/2:F/M). What about when males dominate the groups? Are they as successful as they are when they are alone (66%)? No, from our class data, their intentions are correct only 40% of the time. One solution to this might be that boy or girl dominated groups result in the similarly ineffective results like those results found in unbonded same-sex operator teams of two.

In groups that appear to be male dominated, for example in groups where there is only one female and more than 3 males (i.e. 4, 9, 10, 12) it is evident that the baseline data has a decreased standard deviation compared with the calibration standard deviation of about 7.07.

PSY/ORF Spring 2005 2 Analysis:

What does the Chi-Square test tell us about our class data? In general, a chi-square test is a method of showing the departure of one’s data from the mean. If the Chi^2 is larger than the degrees of freedom, then there is evidence of more departure from the mean. When the Chi^2 is smaller, there is is evidence of less departure from the mean.

According to the previous work done at PEAR, our Low data is right in line with the concept that low intentions lead to more deviation from the mean. Because the devation is squared though,

PSY/ORF Spring 2005 Rescaling Analysis:

In Table 5.0 we have starred all the instances where the graphs rescaled. This generally happened when the mean went above 101.2 and below 98.7. Normally, and by normal we mean in past PEAR experiments, we’d expect the graph to rescale approximately 5% of the time. In our data, however, there are 12 instances of rescaling with another 5 to 8 instances of it coming very close to rescaling. In 72 runs, this is around a 20% rescaling rate which is much higher than expected.