Previous studies showed that observationally acquired stimulus-response (SR) binding and retrieval effects only occur when the observed person is socially relevant. This can, for example, result from task demands (e.g. cooperation or competition, Giesen et al., 2014) or the relationship between the interacting people (Giesen et al., 2018). An essential component of successful social relationships, which could therefore also increase the social relevance of another person, is trust (Lewis & Weigert, 1985). In the present study we investigate whether trusting vs. mistrusting the person one is interacting with modulates observationally acquired SR binding and retrieval effects. Trust was manipulated by a variation of the Investment Game (Berg et al., 1995). Interaction partners behaved either trustworthy or untrustworthy by keeping vs. violating a previous promise. After that, participants performed an online interactive color classification task to assess observationally acquired SR retrieval effects. Experiment 1 showed no modulation of these effects by the level of trust. A possible reason for these results is that many participants indicated that they did not believe to be interacting with another human. After their exclusion, descriptively smaller effects were found in the low trust condition, suggesting that retrieval breaks down for untrustworthy interaction partners. However, this was not statistically significant. In Experiment 2, modifications have been made to make the online interactions appear more realistic. Currently, data is being collected for this modified version. Results will be presented and discussed.