Horses for Courses: Testing DS for psi ability

Susan J. Blackmore and Nicholas Rose
Department of Psychology
University of the West of England
St Matthias College
Bristol BS16 2JP

Paper presented at the 24th International Conference of the Society for Psychical Research, Northampton, September 2000


During the summer of 1998 we were approached by DS who claimed to have psychic ability and to be able to predict the outcome of horse races. DS was confident he could demonstrate that ability in experiments and we were keen to test him. Over a period of nearly two years we carried out a series of experiments to test his claims. At each stage he failed to demonstrate psi and we designed further experiments to take account of his own suggested reasons for failure.

Experiment 1.

DS chose to try guessing the suit of playing cards. In a simple pilot study, carried out in October 1998, SB randomly selected a card once a week and placed it in a location agreed with DS in the office at UWE. DS then phoned in with his guess and NR recorded the guess. Over six weeks DS correctly guessed the suit 3 times in six guesses. He guessed the number five times but was never correct. He said he could not make more frequent guesses but might do better if he had a list of words to look at, as he does in a betting shop.

Experiment 2.

DS and SB created three list of words. Each week SB randomly chose one word from each list and displayed it in her office at home. DS rang in and NR recorded his guesses. 18 targets were prepared. DS made guesses for 14 of them. None was correct. DS and NR discussed how the experiment might be made more like the horse betting situation DS is used to. They decided that a computer programme would be used to simulate a horse race.

Experiment 3.

In a simulated race between 5 horses DS was asked to predict the winner, making his guesses at home. NR ran one trial a week and DS rang in with his prediction at a regular time. When DS rang the race was run and the result recorded by NR. 10 trials were run between 22/4 and 22/6 1999. DS obtained 1 hit (MCE = 2).

Experiment 4.

Conversations with DS revealed that he was unhappy with ringing in his predictions from home and with other aspects of the experiment. NR redesigned the programme to meet the requirements and DS came into the University for further trials. In these experiments DS had to choose the winner from 10 ‘horses’, completing 100 trials over a period of 10 weeks. The computer program recorded predictions and winner. NR recorded the rank of the predicted winner. In the 100 trials DS scored 6 hits (MCE = 10) and a sum of ranks of 578 (MCE = 550).

Experiment 5.

DS expressed concern over the pressure to make 10 predictions in a session, explaining that in the betting shop he sometimes let races run in order to get a feel for the race. NR therefore redesigned the computer programme to allow races to be run without a prediction, and told DS he could make fewer than 10 predictions or even none on a particular session. DS also suggested that stress in his daily life might be affecting his performance on the task. 20 trials were run over three sessions. The computer recorded all the data. DS scored 1 hit (MCE = 2) and a sum of ranks of 106 (MCE = 110). DS’s claims about his performance and his reasons for failure became steadily more complex. SB therefore interviewed DS after this experiment, and the subsequent one. The interviews were tape recorded and transcribed.

Experiment 6.

From the interview DS still appeared confident that he could produce psi if environmental factors at home (such as good sleep and diet) could be improved, and if he had more motivation to win in each individual trial. He also studied his previous results in great detail and on this basis suggested to NR that he would get more 1, 2, 3, 4, 9, and 10 places and fewer 5, 6, 7 and 8’s. NR redesigned the programme to record that information from each race. In addition DS was provided with toy money to make ‘bets’ with. For the 20 trials DS was given £40, allowing him to make an average bet of £2 per race – he could raise or lower that according to his confidence in the outcome. This (in theory) allowed us to take a measure of his confidence, as well as providing him with greater motivation to succeed. 20 trials were run over 6 sessions. DS scored 3 hits (MCE = 2) and a sum of ranks of 99 (MCE = 110). These results, although in the right direction, were not significant. There was no evidence that fewer 5, 6, 7 or 8 places were obtained. DS bet precisely £1 on each race for which he made a prediction, and there was no sign that he was more confident of his prediction when he won. The return of the stake was 7:1 which (with the return of the original stake) meant that he received £8 back for every £1 gambled. From the initial stake of £40, DS ended with a small winnings with a total of £44.

Perhaps the most interesting part of this process was the extent to which DS’s explanation for failure to obtain psi became more complex over time and eventually overcame our ability to simulate features within the experiment. Whilst DS was unable to obtain results suggestive of psi, the results do gradually appear to be moving in that direction – and he did make a small winning in the final experiment. A further experiment is planned if DS is available later in the year.

Thanks to the Perrott-Warrick Fund for support of this project.