Direct mental interactions with living systems (DMILS) research involves testing procedures where a person (an ``agent'') is trying to interact with a biological target system, e.g., another person's physiological responses or the behaviour of small animals or fish. In DMILS studies the biological target is located in a sensorially shielded room, providing isolation from any physical contact with the agent. The target's spontaneously fluctuating activity is monitored continuously while the agent, during randomly interspersed influence and noninfluence (control) periods, tries to influence mentally the target's activity in a pre-specified manner. The target system is unaware of timing or goal orientation (i.e., influence or non-influence) of the agent's mental intentions. When human physiological responses are the target system, the target person's only goal during the experimental session is to remain passively alert and to wish mentally that their physiology will unconsciously respond appropriately to the agent's intentions. The mental strategies used by the agent to interact with the remote, shielded target includes wishing and willing the desired changes to manifest in the target, mental imaging of the desired outcome, and in some instances simply paying attention to the target system. The randomised order of the influence or non-influence period is usually conveyed to the agent by a message on a computer monitor; the monitor may also convey to the agent the actual recordings of the target's activity, thereby providing on-going feedback about the effects of their mental intentions upon the remote target system. The experimental design eliminates possible confounding factors such as recording errors, placebo effects, confounding internal rhythms and chance correspondences.
The majority of the recent DMILS research has been conducted by Braud and his colleagues, who published a meta-analytic summary of 37 of their experiments (Braud and Schlitz ). This work involved 13 different experimenters and 655 sessions. These 37 studies examined seven different target systems, including electrodermal activity (EDA) with the agent trying to influence the subject's EDA to increase or decrease (i.e., trying to ``calm'' or ``activate'' the subject), blood pressure, fish orientation, mammal locomotion, and the rate of haemolysis of human red blood cells. The overall results from this work have been highly significant (per session overall ).
While this work was conducted by 13 different experimenters, it was all performed at the same laboratory. Other laboratories are now attempting to replicate this work, with the initial results generally conforming to those obtained by Braud et al. For example, Delanoy and Sah (1994 ) compared EDA responses to conscious responses in a DMILS environment, in which the agent was either remembering and trying to re-experience a very positive, exhilarating emotion (``activate'' condition) or was thinking of an emotionally neutral object (``control'' condition). The subject's EDA showed significantly greater activity during the activate periods than during the control periods ( . However, the subject's conscious responses (i.e., their guesses as to whether the agent was trying to activate or calm them) did not differ from chance expectancy. The finding of a significant physiological effect, with no corresponding effect shown by a conscious response measure, supports similar findings from Tart  and Targ and Puthoff , and suggests that subtle psi interactions may occur without any conscious recognition on the part of the subject.