because humans attempt to respond by looking whether or not they have vision. There are numerous other inputs that inform us of the presence of others
Essentially Sheldrake evaluates his own work and others pull down his dubious analysis.
I propose the following protocol.
Sometimes that room is otherwise empty, sometimes there are 3 or 4 people (the secondaries). These people are placed in chairs facing away from the location of the subject and which have motors to rotate them. In some experimental runs none of the other 4 persons are rotated. In others rotation occurs randomly and the secondaries are to stare at the subject until the chair rotates away.
After a set period of time, identical in all cases, the primary is taken to a room to be interviewed by someone unaware of the experimental set up for that run. In all cases the video, interview, encephalographic and myographic data are analysed by a third party unaware of the state of the secondaries. The evaluation of the data should look for evidence of voluntary or involuntary movement from the primary and any "feelings" the primary may express about a particular test run.
This analysis is then compared to the data on the presence or absence of staring.
This test has to be carried out on a minimum of 5 times on each of 20 primary subjects with secondaries drawn at random from a pool of at least 100.
Precautions; due to the sensitivity of the human body to floor born vibrations the floor under the Primary should be independent from the floor beneath the secondaries. It might also be advisable to control for body heat and osmic (smell) effects
If you get any more than random noise from such a set-up I would be willing to reconsider my position.