20

Aramburu, Kim, Yu

The Effects of Gender and Method of Persuasion on

Random Processes

PSY/ORF 322

Prof. Jahn

May 9, 2005

Jason Aramburu

Jin-Hee Kim

Lia Yu

“If scientific reasoning were limited to the logical processes of arithmetic, we should not get very far in our understanding of the physical world. One might as well attempt to grasp the game of poker entirely by the use of the mathematics of probability.”

The above quote by Vannevar Bush, the visionary who predicted the rise of the personal computer, illustrates one of the more profound, and often forgotten tenets of modern science: the fluidity of models. In the face of significant and replicable data, extant scientific models sometimes require revision. Take, for instance, the discovery of the structure of DNA by Watson and Crick. Their work, based on careful scientific data, ultimately shattered the notion that proteins were the conveyors of genetic information in living organisms. In addition, it led to the creation of an entirely new model of heredity and cellular replication, which, today, is generally accepted as gospel. However, as scientific models and explanations of reality have become more popularly received, they have also become increasingly rigid. If a sound and replicable study were published that significantly refuted Watson and Crick’s ‘Central Dogma of Molecular Biology,’ it would likely be met with much resistance, both in the scientific community and the public at large. Not only would this hypothetical work alter scientific beliefs, but it would also severely affect people’s perceptions of reality. Although the world has yet to witness such a study, much research has been done to explore the role of consciousness in human-machine interaction. The results of these experiments are equally Earth shattering, and have, so far, been met with similar resistance.

In the 1987 book Margins of Reality, by Robert G. Jahn and Brenda J. Dunne, the results of several years of studies on the influence of human intention on the output of a Random Event Generator (REG) were published. These studies provided statistically significant and reproducible data, which seemed to indicate that human intentions alone could alter the mean output of an REG over a large enough number of trials (Margins, 144). Furthermore, these studies made a profound statement about existing physical models of the relationship between man and machine. Obviously, there must be some unknown avenue by which the intentions of the human ‘operator’ are communicated to the machine.

In 1986, scientist René Peoc’h published a study that further explored the existence of this strange correlation between volition and the output of a random process. In the experiment, young chicks were hatched and raised in the presence of a “Tychoscope,” a small robot capable of making REG controlled movements. The chicks quickly imprinted on the Tychoscope, adopting it as their mother, and were subsequently placed in a transparent cage adjacent to the robot (Peoc’h, 223). The robot was allowed to move randomly through a rectangular area, in full view of the cage. Analysis of the robot’s motion revealed that it spent 2.5 times longer on the half of the rectangle nearest to the chicks, as compared to when it was allowed to run in the presence of an empty cage. Furthermore, statistical analysis of these results yielded p-values less than 0.001, indicating a high degree of statistical significance and deviation from chance (Peoc’h, 224). Graphical results of one such trial are displayed in appendix A.

In a subsequent 1995 study, Peoc’h et al expanded their previous research in order to more clearly isolate the intentions and desires of the chicks (Peoc’h, 226). While research has documented that young birds generally desire to be close to their mother, it is also commonly known that they desire the presence of light when they are awake. Researchers capitalized on this base desire and attached a lit candle to the top of the Tychoscope. The robot was allowed to run in the presence of a cage of chicks that had never seen it before, and thus had not adopted it as their mother. The lights in the room were darkened, so that the chicks could only see the candle on top of the robot. It was assumed that since the burning candle was the only source of light in the room, the chicks would want it to come closer, thereby better illuminating their cage. Data analysis revealed that in 71% of trials, the robot spent most of its time on the side closest to the chicks, whereas its motion was generally random when the chicks were not present. P-values were determined to be less than 0.01, again indicating a high degree of significance (Peoc’h, 226).

The Peoc’h experiments indicate a high degree of correlation between the output of a random process, and the ‘perceived’ intention of a collective operator. The operators’ intentions were merely ‘perceived’ because, obviously, baby chicks cannot relate their true intentions to the experimenters. Furthermore, experimenters had no way of determining if the chicks were actually expressing these perceived intentions. However, several studies have been completed by Jahn et al that indicate a similar correlation between the intention of a human operator and the movement of an REG controlled robot. In these studies, operators were told what their intention should be. In turn, the operators often expressed their intentions by trying to persuade the robot to move a certain way, sometimes using verbal commands or gestures (Dunne, lecture).

The influence of gender on an operator’s ability to significantly shift the mean result of a series of REG trials has been investigated thoroughly. According to a 1998 study by Brenda J. Dunne, there is a correlation between gender and operator success in REG trials. In the study, data from 135 independent operators, gathered between 1979 and 1993,was analyzed along gender lines. It was found that men tended to be more successful, overall, at inducing a mean shift in the output of REG data. In addition, the data indicated females were not as successful at inducing a correlation between their low-output intentions and REG results (Dunne, 1998).

While there are many possible explanations for this gender disparity, only two of the most reasonable will be discussed. One could attribute the decreased success of females to biological differences. While completely unproven, it is possible that males are simply imbued with a more developed ability to form some sort of communicative bond with these REG systems. However, this theory is ultimately problematic, as it is exceedingly difficult to prove, given lack of information on what causes these observed effects. An alternative, and far more testable, theory questions whether the decreased success of females in REG tests is due to their approach, rather than their biological makeup. According to traditional expectations and beliefs, women tend to have more passive and thoughtful personalities than men. Whereas long-standing gender roles assume men to be proactive, alpha-males, women are generally regarded more as organizers, who carefully and deliberately approach and solve problems. Could these traditional stereotypes affect an operator’s style when attempting to induce a mean shift in the output of an REG? Such is one of the fundamental questions of this experiment.

According to a 1997 paper published by R.G. Jahn et al, detailing the results of 12 years of REG studies, insufficient research has been done to determine if the use of a particular operator style is more effective than another (Jahn 1997, 359). Due to the infinite array of potential styles, little systematic research has been performed in this area. However, there does appear to be a correlation between success in REG trials, and an operator’s ability to form some sort of anthropomorphic bond with the device. According to the study, the most successful operators relate to the device as an extension of themselves, in much the same way that a tennis player might think of his racquet as an extension of his arm (Jahn 1997, 359). Could the formation of this bond be a result of an operator’s style?

The purpose of this project is ultimately twofold. First, it aims to determine if an operator’s style, specifically his method of persuasion, has any effect on his ability to influence the movement of an REG controlled robot. Based on the studies performed investigating the role of gender in REG success, two modes of persuasion have been chosen that emulate traditional gender stereotypes. These are the ‘command’ mode, wherein operators are instructed to influence the robot through authoritative demands and orders, and the ‘coax’ mode, wherein operators are instructed to influence the robot through gentle urges and flattery. An operator’s success will be measured in the ‘Time of flight’ of the robot, or the amount of time it takes the robot to fall off of a circular table. The experiment also aims to determine if an operator’s gender has any effect on her performance when using these two modes of persuasion.

It is hypothesized that the traditionally masculine, ‘command’ method of persuasion will be more successful, overall, at inducing a deviation in the mean ‘time of flight’ of the robot. This hypothesis is based on the data that indicates males are more successful in REG studies (Dunne, 1998). Thus, if operators are instructed to adopt this traditionally male style, it is thought that they will be more successful. Similarly, it is believed that when the gender of the operator is examined, the data will indicate that males are more effective at altering the time of flight of the robot when using the command method, while females are more successful when using the coax method. Again, this hypothesis is based on engrained societal norms that associate a commanding personality with masculinity and a coaxing personality with femininity, coupled with the aforementioned REG data (Dunne, 1998).

We tested our hypothesis by means of machinery provided by the Princeton Engineering Anomalies Research (PEAR) laboratory. The specific device which we used was a preexisting program employed in what researches at the PEAR lab refer to as the “Frog Experiment”. The device used by this program is a small robot similar to an electronic car. The apparatus is a vehicle mobilized by two wheels and is capable of spinning around in full 360 degree angles (or more). It moves only forward in the direction it is facing, from 0 degrees to approximately 10 degrees. The robot is situated on a large circular table, 48 inches in diameter, on which it is free range of motion. The angles in which it spins as well as the distance it moves forward are the effect of randomized commands dictated by its onboard Random Event Generator (REG).

On the front portion of the vehicle is a red light. A camera overhead observes the movement of the robot by tracking the movement of the light. The computer program runs constantly throughout the robot’s run gathering information from the overhead camera. The extent of information gathered by the computer from the camera includes the duration of the run, the movement in x and y coordinates, and the angle at which the robot leaves the table. For the purposes of this experiment, however, we only collected data having to do with the duration of each run. When the robot moves out of the designated borders on the table (about an inch and a half in from the edge of the table), the program terminates and displays the information including the duration of the robot’s run, which, by our data, lasts anywhere from 10 seconds to 4 minutes.

Each trial consists of one of this type of run, and each set is a group of 2 trials. We used this program on six participants in total, which the PEAR lab refers to as our ‘human operators’. The makeup of these human operators consisted of both genders, three female and three male. Each participant was responsible for completing 20 sets of data (40 trials). The 20 sets were performed consecutively one after the other. The participants were given instructions on how to operate the machinery and their duty for each trial. The technical aspect of the experiment involved entering the ‘rcam’ operator number (ranging from #790-795) onto the computer database, which enabled the computer to log in their trial data organized under their specific operator number. Each time a new trial began, the human operator had to prompt the program to start recording the new trial by inputting the word ‘rcam’ plus their operator number. After entering this information, the computer would ask them to turn on the robot by flicking a switch on the back of the vehicle.

The computer would log errors if the camera observed that the vehicle was not centered at the start of the run (centering it included alignment with lines of designation and facing a particular direction every trial). Errors would also occur if the camera observed movement before the trial began, if there was movement during the trial by the operator within the camera’s field of vision, or sometimes if there were internal unknown errors which resulted in an alert concerning detected ‘brightness’. At one point during the trials a male operator encountered an error which is worth noting. For this particular operator, operator number 790, trial number 21 was terminated due to a “malfunction” in the machinery, which PEAR lab staff had never encountered before. A full explanation and analysis of this malfunction will be given after more background detail is laid out. If errors were encountered during the trial, the computer would erase the faulty data and restart the trial after the operator prompted it to start again. At the end of each trial, the operator would flick the off switch on the robot, and replace it back in the center of the table. He/she would then record the duration of the run and start the next trial.