Rats were runway trained on three-trial series of different reward events,
plain pellets (P), sucrose pellets (S), and nonreward (N). The series were
constructed so that the position of the N event was always a valid cue for
N. The relative validity of the P and S events, and their respective memori
es as cues, was manipulated by holding the reward kind constant, PSN or SPN
(Experiment 1) and SNS and PNS, or SNP and PNP (Experiment 2), or by varyi
ng the reward kind from series to series (PSN, SSN, PPN, and SPN; Experimen
t 1). In both experiments, transfer tests to series with equivalent numbers
of nonreward trials (NNN) or reward trials (SSS or PPP) showed that the pa
ttern of approach, fast on P and S trials and slow on N, that developed und
er all training conditions was retained in the transfer tests. The transfer
results implicate control by position cues under conditions that might oth
erwise have been interpreted only by reference to reward memories. (C) 2000
Academic Press.