A practice-specific stimulation experiment requires candidate montages that differ in regional electric-field exposure without sacrificing interpretable target dose. We evaluated this requirement for transcranial direct current stimulation (tDCS) montages intended to test focused-attention and open-monitoring mindfulness practices. Four focused-attention candidates and six open-monitoring candidates formed 24 pairs, evaluated with SimNIBS 4.6 in 20 existing anatomical head models at a linearly scaled current of 2 mA. Outcomes were regional 99th-percentile field magnitude, a normalized double-dissociation index (DDI), suprathreshold overlap, and individual dose coverage. Eighteen pairs passed four recorded progression criteria, but only five also met a post hoc requirement that both intended regional contrasts be positive in at least 80% of heads. A positive DDI alone concealed a reversed salience-region contrast in one conventional comparison. The nominated F3-centered 4-by-1 ring paired with F8-Fp2 sponges (FA-05 x OM-02) produced both directional preferences in 20/20 heads, with median DDI 1.561 and median suprathreshold Dice 0.000. The conventional F3-Fp2 and FT8-FCz sponge pair (FA-02 x OM-07) produced both preferences in 19/20 heads, with DDI 0.627 and Dice 0.186. Both anatomical targets reached the operational 0.20 V/m floor in 11/20 versus 20/20 heads, respectively. ROAST 4.0, with ring placement held fixed, supported both nominations but exposed dose disagreement. Across conductivity sensitivity draws, directional separability persisted more consistently than absolute dose. These results establish a regional separability-dose trade-off within the tested candidate set and show why a summed contrast cannot substitute for a directional conjunction. They support selection of a discriminating experiment, not whole-network selectivity, physiological efficacy, or a validated treatment.