Predictably harsh environment is associated with reduced cognitive flexibility in wild food-caching mountain chickadees

Abstract

Cognition is one of the mechanisms underlying behavioral flexibility, but flexibility of cognition itself may vary as a result of trade-offs between the ability to learn new information and the ability to retain old memories. How and when cognitive flexibility is constrained by this trade-off remains poorly understood. We investigated cognitive flexibility in wild food-caching mountain chickadees in the field at different elevations experiencing different levels of environmental harshness, using a spatial learning and memory reversal paradigm. There were no significant differences in sampling strategies between elevations, but high elevation chickadees performed worse than low elevation chickadees on the reversal task, indicating lower cognitive flexibility. Compared to the initial learning task, low elevation chickadees improved their performance during the reversal task, while high elevation chickadees performed worse. High elevation birds inspected previously rewarding locations more frequently than other locations, suggesting that reduced cognitive flexibility is associated with proactive interference. Considering that high elevation chickadees cache more food and are likely more dependent on these caches than their conspecifics from low elevation, and that chickadees from both elevations use similar sampling strategies, our findings suggest that stronger memories of more caches might interfere with acquisition and retention of new memories. Overall, our results suggest that predictably harsh environments might favor stronger memories at the expense of decreased cognitive flexibility, which is likely driven by increased proactive interference.

Natural Selection and Spatial Cognition in Wild Food-Caching Mountain Chickadees

“…We developed and implemented a radio frequency identification (RFID)-based system to test spatial cognition in wild, passive integrated transponder (PIT)-tagged chickadees [21][22][23]. RFIDenabled feeders can be programmed to provide food only to specific individuals while recording the PIT-tag ID and time of each visit by all tagged birds.…”

“…These comparisons were conceived to reveal the presence of selection on two spatial cognition metrics-spatial learning and memory ability and reversal spatial learning and memory ability (when each bird was re-assigned to a new and different rewarding feeder immediately following the spatial learning task) [22,23]. Spatial learning and memory ability are expected to reflect the ability to remember and recover previously made food caches [2,8], while reversal learning ability is expected to reflect learning and memory flexibility, which may be important for foraging success in unpredictable environments [23,25].…”

“…Spatial learning and memory ability are expected to reflect the ability to remember and recover previously made food caches [2,8], while reversal learning ability is expected to reflect learning and memory flexibility, which may be important for foraging success in unpredictable environments [23,25]. We used mean number of location errors per trial over the first 20 trials (initial learning and memory phase) and the mean number of location errors per trial over the entire 4-day task (overall task performance measuring learning and longer-term memory) [21][22][23]. As birds learn, they are expected to make fewer and fewer location errors with each trial; thus, the mean number of location errors per trial over the entire task is a reliable and accurate estimate of learning and memory that is repeatable across years within individuals [21][22][23].…”

“…We used mean number of location errors per trial over the first 20 trials (initial learning and memory phase) and the mean number of location errors per trial over the entire 4-day task (overall task performance measuring learning and longer-term memory) [21][22][23]. As birds learn, they are expected to make fewer and fewer location errors with each trial; thus, the mean number of location errors per trial over the entire task is a reliable and accurate estimate of learning and memory that is repeatable across years within individuals [21][22][23]. Faster learning and better memory are indicated by fewer location errors per trial, so lower mean number of location errors per trial over the entire learning task indicates better learning and memory.…”

“…Faster learning and better memory are indicated by fewer location errors per trial, so lower mean number of location errors per trial over the entire learning task indicates better learning and memory. We used the total number of trials completed during each cognitive task as a covariate in all analyses of cognitive performance to control for individual variation in timing and frequency of trials, which allows testing whether groups differed in cognitive performance independently of potential differences in number of trials [21][22][23].…”