Food-caching mountain chickadees learn abstract rule to solve spatial-temporal task

Cognitive task description: In winter 2021-2022, we tested wild, food-caching mountain chickadees (Poecile gambeli) in their natural environment on a novel spatial-temporal task to test whether chickadees could learn and use an abstract rule while making foraging decisions in a complex spatial-temporal task. In this task, chickadees had to learn to track the daily location of a food reward that moved predictably around a square spatial array of eight feeders. Each day, only one of eight feeder locations provided food to each bird. Each night, the food reward was rotated either clockwise or counterclockwise to the nearest feeder in that direction, so that the following day birds had to find their new rewarding feeder. The food reward was rotated every day in the same direction for each bird for up to 34 days. Hypotheses: The design of this spatial-temporal task allowed us to discriminate between two different processes that birds could use to solve the task. First, if birds relied on associative learning, they would be expected to use memory recency to visit the most recently rewarding feeder, even after the food reward rotated each day. In this case, birds using associative learning and memory should first visit the old rewarding feeder from the previous day, then continue to search by moving backwards to the locations that they remembered recently provided food. Alternatively, if birds learned the rule of the spatial-temporal rotation, they should search for the new rewarding feeder by moving forward toward the correct rewarding feeder and away from the previously rewarding feeders. This second method represents using an abstract rule: birds must learn that the food reward is located one position to the left (or right) of the most recently rewarding feeder. They must apply this learned relationship to new feeders, as the rewarding feeder location continues to rotate around the feeder array each day. We conducted two additional cognitive tasks prior to the spatial-temporal task to assess the role of individual cognitive ability in learning an abstract rule. Using the same spatial feeder apparatus, we tested individual spatial learning and memory ability and single reversal learning ability, often used to estimate cognitive flexibility. We found evidence that chickadees learned and used the abstract direction rule to search for the food reward. This may be the first evidence of abstract rule use in a wild bird in natural conditions.认知任务描述： 2021至2022年冬季，我们在自然生境中对野生储食山雀（Poecile gambeli）开展了一项新颖时空任务测试，以验证山雀能否在复杂的时空觅食决策过程中学习并运用抽象规则。在该任务中，山雀需要学会追踪可预测移动的食物奖励每日位置，该奖励会在由8个喂食器组成的方形空间阵列中变动。每日，8个喂食器中仅有一个会为每只鸟提供食物。每晚，食物奖励会沿顺时针或逆时针方向旋转至该方向上最近的喂食器，因此次日鸟类需找到新的有奖励的喂食器。每只鸟的食物奖励会连续最多34天沿同一方向每日旋转。 研究假设： 本时空任务的设计能够区分鸟类解决该任务可能采用的两种不同认知过程。其一，若鸟类依赖联想学习（associative learning），那么即便每日食物奖励发生旋转，它们仍会利用记忆近期性（memory recency）来访问最近获得过奖励的喂食器。此时，采用联想学习与记忆策略的鸟类会首先访问前一日的旧奖励喂食器，随后通过回溯至近期曾提供食物的位置继续搜寻。其二，若鸟类掌握了时空旋转的抽象规则，则会朝向正确的奖励喂食器向前移动，并远离此前的奖励位置，以此寻找新的奖励喂食器。第二种策略代表了抽象规则的运用：鸟类需要习得食物奖励位于最近获得奖励的喂食器左侧（或右侧）一个位置的规则，并将这一习得的关联应用到新的喂食器位置中，因为每日奖励喂食器的位置都会在喂食器阵列周围旋转。在开展本时空任务之前，我们还开展了两项额外的认知任务，以评估个体认知能力在学习抽象规则过程中的作用。我们使用相同的空间喂食器装置，测试了个体空间学习与记忆能力，以及通常用于评估认知灵活性（cognitive flexibility）的单次反转学习（reversal learning）能力。 我们发现证据表明，山雀学习并运用了抽象方向规则来搜寻食物奖励。这可能是野生鸟类在自然条件下运用抽象规则的首个实证证据。