Invisible attitudes have visible effects on the behaviour of animals in scientific studies.

Animal behaviour research requires measures to avoid human influences on the investigated animals. While efforts to minimize experimenter biases have primarily focused on reducing explicit expectancy effects, the impact of human attitudes on animal behaviour, particularly on the unconscious level, remains largely unstudied in research contexts. Here we show that experienced animal handlers' explicit (conscious) and implicit (unconscious) attitudes influenced the behaviour of similarly raised, human-socialized dogs and wolves during commonly used sociability tests. Explicit attitudes shaped animals' affiliative and discomfort behaviour, while implicit attitudes affected handlers' heart rate variability which, in turn, modulated animal affiliation. Notably, explicit and implicit attitudes were uncorrelated, highlighting that they may have had different effects. While wolves reacted more to human attitude and physiology than dogs, time series analyses revealed that the behavioural influence between both canid species and humans was strong and reciprocal. Our findings highlight the subtle yet powerful role of human conscious and unconscious attitudes in shaping animals' behaviour in research and beyond, potentially creating artifacts and skewing reported differences between species or populations. Our results raise concerns about findings in comparative cognition and behaviour research, emphasizing the need for more appropriate control of attitude-related experimenter bias.