In Experiment 1, 3 college students were exposed to relational pretraining to establish the contextual functions of Same, Opposite, More Than, and Less Than in four arbitrary stimuli. Subjects were then trained on the matching-to-sample tasks A 1-81 and Y1-N1, in the presence of the More-Than contextual cue, A 1-82 and Y1-N2 in the presence of the Less-Than contextual cue, C1-D1 and E1 -D2 in the presence of the Same cue, and C1-D2 and E1-D1 in the presence of the Opposite cue. Test trials were subsequently administered to probe for the mutually entailed relations; Less-Than/81-A 1, Less-ThanlN1-Y1, More-Than/82-A1, More-Than/N2-Y1, Same/D1-C1, Same/D2-E1, Opposite/D2-C1, and Opposite/D1-E1. Response latencies to probes for derived Same/Opposite relations were significantly lower than those for derived More ThaniLess Than relations. Experiment 2 exposed 4 subjects to training across each of the four relations and used a novel stimulus set to test for reduced response latencies to the derived relations. Response latencies to More-ThaniLess-Than probes reduced significantly across the original to the novel stimulus set, whereas latencies to Same/Opposite probes were low across both stimulus sets.