Ojeda Andrés, Murphy Robin A, Kacelnik Alex
Department of Zoology, University of Oxford, OX1 3PS, UK.
Department of Experimental Psychology, University of Oxford, OX2 6GG, UK.
Behav Processes. 2018 Jul;152:73-80. doi: 10.1016/j.beproc.2018.03.024. Epub 2018 Mar 30.
Decision-makers benefit from information only when they can use it to guide behavior. However, recent experiments found that pigeons and starlings value information that they cannot use. Here we show that this paradox is also present in rats, and explore the underlying decision process. Subjects chose between two options that delivered food probabilistically after a fixed delay. In one option ("info"), outcomes (food/no-food) were signaled immediately after choice, whereas in the alternative ("non-info") the outcome was uncertain until the delay lapsed. Rats sacrificed up to 20% potential rewards by preferring the info option, but reversed preference when the cost was 60%. This reversal contrasts with the results found with pigeons and starlings and may reflect species' differences worth of further investigation. Results are consistent with predictions of the Sequential Choice Model (SCM), that proposes that choices are driven by the mechanisms that control action in sequential encounters. As expected from the SCM, latencies to respond in single-option trials predicted preferences in choice trials, and latencies in choice trials were the same or shorter than in single-option trials. We argue that the congruence of results in distant vertebrates probably reflects evolved adaptations to shared fundamental challenges in nature, and that the apparently paradoxical overvaluing of information is not sub-optimal as has been claimed, even though its functional significance is not yet understood.
只有当决策者能够利用信息来指导行为时，信息才对他们有益。然而，最近的实验发现，鸽子和椋鸟重视它们无法利用的信息。在这里，我们表明这种矛盾在大鼠中也存在，并探索其潜在的决策过程。实验对象在两个选项之间进行选择，这两个选项在固定延迟后以概率方式提供食物。在一个选项（“信息”）中，选择后立即给出结果（有食物/无食物），而在另一个选项（“无信息”）中，直到延迟结束结果才确定。大鼠通过选择“信息”选项牺牲了高达20%的潜在奖励，但当成本为60%时，它们会改变偏好。这种逆转与在鸽子和椋鸟身上发现的结果形成对比，可能反映了值得进一步研究的物种差异。结果与序列选择模型（SCM）的预测一致，该模型提出选择是由在连续遭遇中控制行动的机制驱动的。正如SCM所预期的那样，单选项试验中的反应潜伏期预测了选择试验中的偏好，并且选择试验中的潜伏期与单选项试验中的相同或更短。我们认为，在远亲脊椎动物中结果的一致性可能反映了对自然界中共同基本挑战的进化适应，并且尽管信息的功能意义尚未被理解，但对信息明显矛盾的高估并非如人们所声称的那样是次优的。
