Integrated information increases with fitness in the evolution of animats.
Integrated information increases with fitness in the evolution of animats.
复制标题
DOI：
10.1371/journal.pcbi.1002236
复制
发表时间：
2011-10
影响因子：
4.3
通讯作者：
Adami C
中科院分区：
文献类型：
--
作者：
Edlund JA;Chaumont N;Hintze A;Koch C;Tononi G;Adami C
生物有机体的特征之一是它们能够整合不同的信息源，以优化它们在复杂环境中的行为。如何将这种能力量化并将其与生物体的功能复杂性联系起来仍然是一个具有挑战性的问题，特别是因为生物体的功能复杂性并没有明确定义。我们在这里提出了几个候选度量，量化信息和整合，并研究它们对适应度的依赖，因为人工智能体（“动物”）经过数千代的进化，在简单的模拟环境中解决导航任务。我们比较了这些措施预测高适应度的能力与更传统的信息理论处理措施。当动物通过增加对世界的“适应”来适应时，信息整合和处理就会沿着进化的下降线相应地增加。我们认为，适应度与信息整合和加工措施的相关性表明，高适应度既需要信息处理，也需要信息整合，但当任务需要记忆时，信息整合可能是一个更好的衡量标准。信息整合（以及信息处理）和适应度的度量的相关性强烈地表明，这些度量反映了动物的功能复杂性，并且即使在缺乏适应度数据的情况下，这些度量也可以用于量化功能复杂性。智能行为包括在复杂环境中适当的导航，这是通过整合感官信息和对过去事件的记忆来创造有目的的运动来实现的。这种行为通常被描述为“复杂的”，但量化这种概念的通用方法并不存在。测量功能复杂性的有希望的候选方法是基于信息理论，但没有考虑到记忆在复杂导航中所起的重要作用。在这里，我们研究了一种不同的信息论度量，称为“综合信息”，并研究了它反映同时使用感官数据和记忆的导航复杂性的能力。我们认为，当记忆成为导航策略的关键因素时，基于综合信息概念的测量方法比其他标准测量方法与适应度的相关性更好，但如果机器人使用纯粹的反应性传感器-电机回路进行导航，则基于综合信息概念的测量方法与更标准的信息处理测量方法的效果一样好。我们得出的结论是，从感官数据流中发出的信息与过去事件的一些（短期）记忆的整合对于复杂和智能行为至关重要，并推测整合信息——在某种程度上它可以被测量和计算——可能最好地反映动物行为的复杂性，包括人类行为。
One of the hallmarks of biological organisms is their ability to integrate disparate information sources to optimize their behavior in complex environments. How this capability can be quantified and related to the functional complexity of an organism remains a challenging problem, in particular since organismal functional complexity is not well-defined. We present here several candidate measures that quantify information and integration, and study their dependence on fitness as an artificial agent (“animat”) evolves over thousands of generations to solve a navigation task in a simple, simulated environment. We compare the ability of these measures to predict high fitness with more conventional information-theoretic processing measures. As the animat adapts by increasing its “fit” to the world, information integration and processing increase commensurately along the evolutionary line of descent. We suggest that the correlation of fitness with information integration and with processing measures implies that high fitness requires both information processing as well as integration, but that information integration may be a better measure when the task requires memory. A correlation of measures of information integration (but also information processing) and fitness strongly suggests that these measures reflect the functional complexity of the animat, and that such measures can be used to quantify functional complexity even in the absence of fitness data. Intelligent behavior encompasses appropriate navigation in complex environments that is achieved through the integration of sensorial information and memory of past events to create purposeful movement. This behavior is often described as “complex”, but universal ways to quantify such a notion do not exist. Promising candidates for measures of functional complexity are based on information theory, but fail to take into account the important role that memory plays in complex navigation. Here, we study a different information-theoretic measure called “integrated information”, and investigate its ability to reflect the complexity of navigation that uses both sensory data and memory. We suggest that measures based on the integrated-information concept correlate better with fitness than other standard measures when memory evolves as a key element in navigation strategy, but perform as well as more standard information processing measures if the robots navigate using a purely reactive sensor-motor loop. We conclude that the integration of information that emanates from the sensorial data stream with some (short-term) memory of past events is crucial to complex and intelligent behavior and speculate that integrated information–to the extent that it can be measured and computed–might best reflect the complexity of animal behavior, including that of humans.