Phenomenologists, in particular the early Heidegger, suggest to conceive of the human agent1 as being-in-the-world [15]. What is clear from this term by itself is that the world should not be conceived of as something separate from the human being, but as intimately related to what it is to be a human agent. Such a view stands in contrast with an observant, detached 'Cartesian Ego' that mainly contemplates about the world, as Michael Wheeler explains in more detail [16]. Furthermore, the phenomenological tradition conceives of the human being in practical interaction with the environment. According to Merleau-Ponty [17], "[c]onsciousness is in the first place not a matter of 'I think that' but of 'I can'". Notably, the 'I can' implies the possibility of change in the actual world. While the 'I think' refers to reflection about the world, the 'I can' acknowledges the continuous potential for change.
The phenomenological focus on interaction with the environment has resulted in specific terminology, emphasizing that worldly entities appear first of all as tools, as things that enable us to perform certain actions ([18], see also [19]). The objects are engaged in a way that has already appreciated the possibilities, options, or opportunities provided by these objects. Heidegger's famous example is his analysis of our appreciation of a hammer [15]. The hammer is not primarily perceived as a 'thing' with a wooden component and an iron part, but it has already been perceived - before its parts are recognized in their specific nature - as something which enables us to hammer. In everyday life, we engage the 'objects' as enabling certain actions, and therefore, indeed, as tools. Like when we are building, we pick up the tools without much thought; we are engaged in a certain praxis, which provides us with the practical eye that makes us recognize the specific tools suitable for the actions we intend to perform. In sum, we appreciate the world from a perspective that continuously recognizes actual possibilities for action.
The possibilities appreciated by the agent are not limited to one simple option [19]. Rather, we recognize a range or network of options. Returning to the hammer, this tool does not merely refer to the act of hammering; it rather opens up further possibilities, like fixing things and building a shed or a house. In fact, possibilities or options go on infinitely, because every possibility opens up a plethora of further possibilities [16]. In order to acknowledge that it is not about isolated possibilities, but about a range of interdependent options, Wheeler uses the term involvement networks: "...the hammer is involved in an act of hammering; that hammering is involved in making something fast; and that making something fast is involved in protecting the human against bad weather." [16] The analysis so far comes down to a basically action-oriented approach to our being in the world, in which practical options and possibilities provided by the actual encounter with the environment constitute the primary level of our understanding of the world.
Wheeler has integrated Heidegger's phenomenological analyses in the philosophy of artificial intelligence (AI) - in part building on Dreyfus's earlier work [20]. Wheeler is certainly not merely using Heideggerian phenomenological notions. A clear example is the central role of the body in Wheeler's account. In Heidegger's Being and Time, the body is absent, as Heidegger himself ([15], p. 143) says: "This 'bodily nature' hides a whole problematic of its own, though we shall not treat it here." Meanwhile, the body is to be found - and its role emphasized - in the work of other representatives of the phenomenological tradition, like Merleau-Ponty [17], who was one of the sources for Dreyfus's initial criticism of AI [21] (see also [16], p. 167) that inspired Wheeler's account. So, Wheeler's work is informed also by other phenomenological strands (other than Heidegger) enabling him to articulate the embodied nature of human agency.
According to Wheeler, engagement with the world is an ongoing adaptive process with continuous action-oriented perception [16]. He understands the engaged attitude toward the world as a form of 'online intelligence': "A creature displays online intelligence just when it produces a suite of fluid and flexible real-time adaptive responses to incoming sensory stimuli." This formulation shows a basic view of how both organisms (humans included) and robots relate to their environment - and is, indeed, a radical interpretation and application of some Heideggerian notions. Online intelligence is especially relevant in a world that is constantly changing, like our world. It is the opposite of offline intelligence: detached cognitive processes that are not in immediate interaction with the world, like contemplating the weather in Paris [16]. Offline intelligence, in other words, is the opposite of embodied-embedded cognitive activity. More can be said about such a distinction, and how it should be conceived of [22], but in this paper I intend to use it in Wheeler's sense.
For decades, Wheeler explains, people in AI tried to build robots equipped with cognitive maps of the world: these maps represented all kinds of aspects of the world. Contrary to what was expected, such robots were not able to smoothly interact with the environment. More recently, other types of robots emerged. These robots did not know that much about the world, but they were designed to continuously pick up environmental cues while interacting with the world. Without a precise representation of the world (so, without a map), but equipped with the capacity to continuously sense the world and interact with it, they are capable of smooth interaction. For this kind of robots their specific bodily features and abilities to interact are crucial [16]. So, without an elaborate map, but equipped with some relevant sensors, they are able to effectively interact with their environment - in a way their clumsy and detached predecessors were unable to.
In other words, an 'online' approach to the interaction with the world has been helpful to AI and robotics. It highlights that, contrary to what one might think, it is the actual situation that is the enabler of options and behavioral scenarios. In fact, AI for a long time overlooked the specific nature and complexity of interacting with the world. For example, three decades ago, IA specialists thought that the major challenge was building a chess computer able to beat the world champion. But beating a chess grandmaster - even the best of the (human) world - turned out to be the easy part. The real challenge was not this kind of cognitive activity, but real-time, online interaction with the world, like ants, mice, and falcons perform. Already several decades ago, Hubert Dreyfus identified and explained the problems and challenges in AI referring to Heideggerian notions like 'being-in-the-world' [16, 20, 23]. His account brought forward that interaction with the world is something much different from what we theoretically anticipated - still, in practice, we, as humans, are extremely good at it.
As indicated, the concept of online intelligence is intimately related to 'embodied-embedded cognitive science'. Wheeler ([16], p. 11) says, "[i]n its raw form, the embodied-embedded approach revolves around the thought that cognitive science needs to put cognition back in the brain, the brain back in the body, and the body back in the world." In my account I emphasize cognition, body, and the environment - not the brain. More in particular, I take the situated embodied nature of our being in the world to be the core of online intelligence within the context of this paper.
It is well established that the phenomenological tradition emphasizes the role of emotions in our being in the world [19, 24, 25]. This has to be acknowledged when considering profound disturbances of emotions - like in anxiety disorders, GAD being an example - from a phenomenological perspective. Heidegger, who also had a particular interest in psychiatry [26, 27], explains that our engagement with the world is always taking place in some mood [15]. Mood, like the weather, is always there. We may be happy, sad, or anxious, but there is always a mood in which we engage our environment. And our mood is profoundly related to the actual options we appreciate in the environment [15, 25]. It is, therefore, likely that anxious patients will appreciate and perceive all kinds of smaller and bigger threats in their environment because their mood makes them focus on such dangerous possibilities. So, an anxious patient won't have much trouble finding things to worry about and one could phenomenologically explore this issue with respect to GAD. However, this is not the focus of my paper; I take the 'online intelligence' angle on worrying in GAD.
Notably, usually online and offline cognitive activity go hand in hand: we combine these processes. In fact, in our everyday activities it is about a balance between these two, and about the ability to change one's approach in accordance with the task we perform.
In the next section I show how both accounts - concerning the unhelpful worrying strategy on the one hand and the online intelligence view on the other - can be linked in a way that brings forward a deeper metacognition in GAD: a metacognition about the nature of our interaction with the world.