Sensory and working memory in a spatial change-detection task by pigeons and humans.

Judgements of items viewed less than 100 ms prior are predominantly supported by a sensory, or iconic, memory system. Iconic memory is of high-capacity, but is also volatile and limited in duration. Judgements after longer delays increasingly rely on a working memory system, which is lower in capacity and volatility than sensory memory, but is longer in duration. In four experiments, several factors (e.g., length of delay, number of items, time to view items, presence of a visual mask) were manipulated during a spatial change-detection task conducted with humans and pigeons. Both species were exposed to trials with an array of colored circles (2, 3, and 4 circles in Experiment 1 and 2a; 4, 6, and 8 circles in Experiment 2b) followed by a brief delay (0, 50, and 100 ms in Experiment 1a; 0, 100, and 1000 ms in Experiments 1b and 2), and then were presented with a test display in which the position of one of the items had changed. Pigeons, like humans, were less accurate in selecting the changed item with more items in the display and after longer delays. Pigeons were equally accurate on trials with 0 and 100-ms delays, but worse on trials with a 1000-ms delay; whereas, humans were equally accurate on 100-ms and 1000-ms delays, but better on 0-ms delay trials. Accurate change detection was disrupted in both species when a visual mask was inserted between the sample and test display after a short (100 ms), but not a long (1000 ms) delay. The results support similarity between species in the functional relationships between delay and memory systems, despite time course differences related to sensory memory.

The influence of landmark stability on control by occasion setters.

In an operant serial feature-positive procedure, an occasion setter (OSX) signals that a response will be reinforced in the presence of a second stimulus (e.g., a discriminative stimulus, A). During a transfer test, the OS is paired with a different discriminative stimulus. Experiment 1a tested transfer effects in a touchscreen-based spatial occasion setting task with pigeons. During training, four OSs (OSW, OSX, OSY, and OSZ) were paired on separate trials with landmark A (LMA) or B (LMB) and the opportunity for a reinforced response at one location to the immediate left (R1) or right (R2) of the LM (OSW→LMA:R1, OSX→LMA:R2, OSY→LMB:R1, OSZ→LMB:R2). Training also included non-reinforced trials of LMA and LMB alone (LMA- and LMB-) and trials of a non-modulated LM with R1 and R2 reinforced across separate trials (LMC:R1 and LMC:R2). After training, the number and spatial location of responses during test trials of a LM paired with the same OS as in training did not differ reliably from transfer tests of an OS paired with a different, modulated LM (OSW→LMB and OSY→LMA), but did differ from transfer to the non-modulated LM (OSX→LMC). Experiment 1b utilized the same pigeons and training with LMB to test the degree to which the spatial stability of a LM influenced transfer. Retraining with LMA was intended to establish it as a non-modulated, stable LM (LMA:R2). Subsequent tests with LMA revealed reduced modulation by the formerly trained OS (OSW), and complete disruption of modulation of spatial location during transfer with a different OS (OSY). These findings further our understanding of the conditions under which OSs may develop and transfer modulation.