Epistemic Structure: How Agents Change the World for Cognitive Congeniality

A two-part model is presented to explain how some organisms and humans add structure to the world to reduce cognitive complexity. Some philosophical possibilities of this model are then outlined, which include a situated cognition explanation of representation and intentionality. Many organisms generate stable structures in the world to reduce cognitive complexity (minimize search or inference), for themselves, for others, or both. Wood mice (Apodemus sylvaticus) distribute small objects, such as leaves or twigs, as points of reference while foraging. They do this even under laboratory conditions, using plastic discs. Such "waymarking" diminish the likelihood of losing interesting locations during foraging (Stopka & MacDonald, 2003). Red foxes (Vulpes vulpes) use urine to mark food caches they have emptied. This marking acts as a memory aid and helps them avoid unnecessary search (Henry, 1977, reported in Stopka & MacDonald, 2003). The male bower bird builds colorful bowers (nest-like structures), which are used by females to make mating decisions (Zahavi & Zahavi, 1997). Ants drop pheromones to trace a path to a food source. Many mammals mark up their territories. At the most basic level, cells in the immune system use antibodies that bind to attacking microbes, thereby "marking" them. Macrophages use this "marking" to identify and destroy invading microbes. Bacterial colonies use a strategy called "quorum sensing" to know that they have reached critical mass, (to attack, to emit light etc.). This strategy involves individual bacteria secreting molecules known as auto-inducers into the environment. The auto-inducers accumulate in the environment, and when it reaches a threshold, the colony moves into action (Silberman, 2003). Such 'doping' of the world is commonly seen in lower animals, most large animals (large body & brain size) do not exploit this strategy. Except humans. More than any other species, humans generate external structure to reduce cognitive complexity, for themselves and/or for other humans. Markers, color-codes, page numbers, credit-ratings, badges, shelf-talkers, speed bugs, road signs, post-it notes, the list is almost endless. Humans also create external structure for reducing the cognitive complexity of artifacts. Examples include bar codes (help check-out machines' decisions easier), content-based tags in web pages (makes Web agents' decisions easier), sensors on roads (makes the traffic light's decisions easier), etc. The pervasiveness of such structures across species indicates that adding structure to the world is a fundamental cognitive strategy (Kirsh, 1996). Note that these structures predominantly serve a task-smoothening function – they make tasks easier for agents. Some of these structures have referential properties, but they do not exist for the purpose of reference. They exist to make tasks easier. From here onwards I will term such stable structures for "cognitive congeniality" (Kirsh, 1996), epistemic structures. The term is derived from a distinction between epistemic and pragmatic action made by Kirsh (1994). How do organisms generate and use such structures? This is the question I want to address in this paper. I am primarily interested in the human case, but I will explore the case of humans adding structure to the world in the broader context of other organisms exhibiting such behavior. A Taxonomy and a Property Most of the literature on epistemic structures is by David Kirsh, and from the Distributed Cognition tradition in general. Kirsh’s work explores the structural and computational properties of such structures, and how they work. I am interested in the other half of the problem, i.e. how such structures are generated and used. I will use Kirsh's model of how such structures work to develop a model of how such structures are generated. Some philosophical possibilities presented by the model are then outlined, along with future work. Epistemic structures can be classified into three, based on whom they are generated for. Examples in brackets. 1. Structures generated for oneself (Cache marking, bookmarks) 2. Structures generated for oneself and others (Pheromones, color codes) 3. Structures generated exclusively for others (Warning smells, badges) A central feature of such structures is their task-specificity (more broadly, function/goal-orientedness). Here's an example to illustrate the concept. Think of a major soccer match in a large city, and thousands of fans arriving in the city to watch. The organizers put up large soccer balls in every junction, and on streets leading up to the venue. Obviously, the ball reduces the fans' cognitive load, but how? To see how, we have to examine the condition where big soccer balls don't exist to guide the fans. Imagine a soccer fan walking from his hotel to the game venue. She makes iterated queries to the world to find out her world state (What street is this? Which direction am I going?), and then does some internal processing on the information gained through the queries. After every few set of iterated queries and internal processing, she updates her world state and mental state, and this continues until she reaches her destination. What changes when the ball is put up? The existence of the big soccer ball cuts out the iterated queries and internal processing. These are replaced by a single query for the ball, and its confirmation. The agent just queries for the ball, and once a confirmation comes in, updates its world state and internal state. The ball allows the agent to perform in a reactive, or almost-reactive mode, i.e. move from perception to action directly. This happens because the ball is a task-specific structure it exists to direct soccer fans to the game venue. Other structures, like street names and landmarks in a city, are function-neutral or task-neutral structures. The fans have to take these task-neutral structures and synthesize them to get the task-specific output they want. Once the huge ball, a task-specific structure, exists in the world, they can use this structure directly, and cut out all the synthesizing. (How the soccer fans manage to discover the ball's task-specificity is a separate and relevant issue, but we will not address it here.) Task-specificity is a common property of all epistemic structures, including pheromones and markers. Kirsh's model of “changing the world instead of oneself” (Kirsh, 1996), postulates that generation of external structures involve task-external actions, and the structures work by deforming the state space in a way that paths in a task environment are shortened. Such structures also allow new paths to be formed. Kirsh’s model only tackles physical structures generated by organisms, like tools. He does not consider structures generated for cognitive congeniality. In the above example, I treat the task environment as a combination of the agent and the environment (what I term action-environment), a notion Kirsh resists. He thinks such a view devalues the generality of the notion of task environment. I argue below that this need not be the case.