Towards a statistical approach to humanistic-geographical place concepts

Abstract

The long-standing division between quantitative and qualitative approaches to studying place remains a challenge. This article seeks to reconcile both viewpoints from a humanistic perspective and focusing on statistics. It outlines a humanistic concept of place compatible with statistical approaches, locates randomness in place, and discusses two key elements for a prospective statistical school of thought on place: an attitude with which place can be adequately approached statistically and proposals for stochastic processes concerning place. The goal is to open up the prospect of a future productive statistical methodology for place that avoids the otherwise so common critiques.

Introduction

“Alice opened the door and found that it led into a small passage, not much larger than a rat-hole: she knelt down and looked along the passage into the loveliest garden you ever saw. How she longed to get out of that dark hall, and wander about among those beds of bright flowers and those cool fountains, […].” - Lewis Carroll (1865: 8–9)

Place has been notoriously difficult to formalize. We constantly construct and use places and are very familiar with the concept at an intuitive level: we typically have a home (not just a location on the map), a workplace (which is often more than just the location where we earn our living), a favorite bench in the park (that is unlike all those other benches despite offering the same functionality), and so on. Yet, it is not easy to put such places appropriately into more formal words. A representation of the garden Alice is confronted with in the above quotation from Lewis Carroll’s “Alice in Wonderland” by means of a polygonal geometry with the attribute green and overlaid with two GIS layers to represent the described flowerbeds and fountains would not do justice to the impression the work evokes when read. Just as in everyday life, when we concretely experience places to which we ascribe meanings, the above quote has the capacity to communicate to us a rich world of experience and meaning. This world contains emotional components, but also a described material context, which makes this emotional relation to the places in question possible in the first place. This form of engagement with meaning-laden places has traditionally been the subject of humanistic geography, which has enjoyed particular popularity since the late 1950s and has seen an increasing interest, again, in recent years. Humanistic geography favors phenomenological forms of inquiry and thus expansive methods (Seamon and Lundberg, 2017). It thus resists atomistic approaches such as those commonly found in GIS, or reductionisms like in statistical analysis. As a result of the rejection of these forms of inquiry, a separation has developed between those who are offspring of the quantitative revolution (e.g., geographical information scientists) and those who reject the accompanying reductionist approach focused on revealing regularities and laws as inappropriate to place as an object of inquiry (Gyuris and Michel, 2024). In this article, I aim to challenge this schism by offering possible ways in which a quantitative approach in the sense of statistics may find its way into the methodological repertoire of humanistic geography. I will also discuss what role stochasticity plays in the place concept underlying the humanistic school.

Place has recently gained prominence as a topic also beyond human geography, including in geographical information science. This is reflected in a number of events such as the ongoing PLATIAL symposium series (Mocnik and Westerholt, 2020, 2022; Westerholt et al., 2018; Westerholt and Mocnik, 2023) and several international workshops on pertinent topics (e.g., Scheider and Adams, 2013; Vasardani et al., 2012; Winter et al., 2009). A comparatively early systematic review on the topic and from a geographical information science (GIScience) perspective was conducted by Merschdorf and Blaschke (2018). They found that individual facets of place are considered in subfields such as critical and participatory GIS, geosemantics, and toponym research. However, they also state that a more systematic and holistic approach to place integrating individual facets is still missing. Rather, different subfields follow their own interests and are influenced to varying degrees by other disciplines such as psychology, philosophy, and human geography. Hamzei et al. (2020) undertook another systematic review to compile an inventory of the types of place facets addressed in the literature. The main facets they identified include anthropocentric (e.g., affordance and sense of place), linguistic (e.g., toponyms and qualitative spatial relations), and so-called derived facets (e.g., symbolic representation and identity). Tang and Painho (2021) and Acedo et al. (2022) have extended this facet perspective with a systematic exploration of concrete operationalizations and identifying relevant research questions. They identify a number of reductionist approaches, including user-generated geographic information, scale operationalizations, and uncertainty quantification, to name but a few. This shows that GIScience is not only disjointed in relation to place, as Merschdorf and Blaschke (2018) note, but that the field also adopts a reductionist perspective that mostly considers individual aspects of place in isolation. Also, many of the above examples follow the established conceptual lines of spatial science (see Kitchin, 2015) and only a few formal constructs have been developed specifically for place.

Initial ideas and proposals for a more systematic approach to the concept of place in GIScience exist in limited numbers. Scheider and Janowicz (2014) propose a model for place references. They argue that GIS only provides ways to refer to locations in a geometric, geospatial sense, and propose a model for locating and identifying places based on affordances and simulated activities. Papadakis et al. (2016), Papadakis and Blaschke (2017), and Papadakis et al. (2020) relax the focus on affordances and offer a function-oriented view of place. They argue that this better reflects how a place operates, rather than how someone experiences what the place may only offer to a single subject. In this way, the authors argue, their approach is less tied to subjects and thus more intersubjective. Purves et al. (2019) present an ontology of place based on what are called the core concepts of spatial information (see Kuhn, 2012). Using the elements of place, field, object, network, and event, the authors posit that “[a] place is an object resulting from a shared identification of a location. As an object, it may become a part of a network and participate in events” (Purves et al., 2019: 1176). The approaches outlined have merit in the context of information systems. However, they often do not explicitly address some of the key aspects of the various schools of thought of place that exist in human geography, but incorporate concepts from GIScience, spatial cognition, psychology, and others. Accordingly, Wagner et al. (2020), in a systematic review, found that many GIScience scholars engaged in the study of place do refer to human-geographic conceptions of place (mostly to the humanistic literature), but tend to mix different conceptualizations that may not be commensurable. Mocnik (2022) also notes this and goes a step further by not providing another ontology of place, but instead suggesting possible paths towards what he calls “platial information science.” He proposes to do this in dialogue with geographical (and other) concepts of place (e.g., from cognitive science and linguistics), arguing that any theory of platial ¹ information inevitably requires a rooting in some notion that relates to actual places (and not just information about them). However, Mocnik (2022) also repeatedly refers to the property of idiosyncrasy that he ascribes to many places. This property may limit a nomothetic treatment of place, such as that pursued in the remainder of this paper, showing that a thorough discussion of place and statistics is a much needed exercise.

The interface between place and statistics has not yet been comprehensively and systematically explored. In fields such as psychology and cognitive science, it is indeed common practice to analyze elements of place statistically, utilizing methods such as mental mapping or surveys as forms of data collection, and with regression, factor analysis, and structural equation modeling to analyze responses. Examples of such approaches to quantifying individual elements of place from recent decades include work such as developing a scale to measure place identification (Burdge and Ludtke, 1972), a scale to determine the degree of placelessness or place affinity of respondents (Shamai, 1991), measures of “feeling at home” (Cuba and Hummon, 1993) and “rootedness” (McAndrew, 1998), to name but a few. Multidimensional approaches have also been developed to capture different elements of a place and subject them to statistical analysis. For example, Williams et al. (1992) offer a range of operationalizations that capture both place identity and place dependence. More recently (and now a frequently used conceptualization in environmental psychology), a model offered by Jorgensen and Stedman (2001) allows sense of place to be broken down into place attachment, place identity, and place dependence, each of which can be operationalized through a number of features. Recent examples for the utilization of the latter include analyses of sense of place in relation to urban amenities in Lisbon, using structural equation and logistic regression modeling (Westerholt and Acedo, 2024; Westerholt et al., 2022) and the work of McCunn et al. (2023), which explores the relationship between sense of place and connection to nature. The work I have highlighted in my brief walk through a few decades falls within the field of psychometrics, which is concerned with the measurement of latent psychological constructs (Rust and Golombok, 2014). However, the concept of place as used in geography goes beyond psychological aspects as “[p]lace is territorial, social, cultural and psychological” (Ilovan and Markuszewska, 2022: 4) and is moreover characterized by the aspect of wholeness. In psychometrics, instead, attempts are often made to dissect place and focus on specific parts of the concept in order to isolate as clearly as possible relevant associations. This is legitimate and also productive, but somewhat limited for the purpose of bringing together GIScience and statistics in a place context since the former is concerned with geographical information and therefore, I argue, a geographical conceptualization of place.

“Humanism has variously impacted upon geographic practice […] as in large part a reaction against positivist social science” Rodaway (2015: 334). This quote reflects well on why I intend to contribute to closing the gap between human geography and spatial science: the emergence of humanistic positions in geography was inextricably linked to the rejection by some of a growing interest in formal, quantitative methods. The remainder of this paper contributes to bridging those positions and offers an epistemological and methodological discussion of possible paths toward what, in reference to “platial information science” (Mocnik, 2022) and analogous to the term “spatial analysis,” could be referred to as “platial analysis” (or rather not analysis, as we shall see later in our discussion). As pointed out in the previous paragraphs, place has so far often been reduced to attributes attached to spatial references (Scheider and Janowicz, 2014) and is thus often reduced to “locations with meaning” in GIScience, where the term “location” is crucial because it refers to geometric space. This view offers a reductionist approach and does not always allow a clear conceptual demarcation from the conceptually different notion of “point of interest,” which refers to third places beyond home and work denoting “interesting or relevant named place[s]” (Psyllidis et al., 2022: 1). Some scholars have taken this up, for example, Quesnot and Roche (2015), who note that social media data is platial rather than spatial; or Gao et al. (2013), who propose platial buffers and joins. From the perspective of spatial analysis and in the context of analyzing spatial footprints of places, Westerholt (2019) outlines some challenges that would need to be tackled for true platial analysis. These include more appropriate formal constructs and units for representing places, and adapted methods such as estimators of platial rather than spatial autocorrelation (among other estimable properties). I maintain that many of these steps should be postponed for now, as we first need to discuss some of the more fundamental aspects of the interface between place and statistics. Westerholt (2019) also argues that a discussion of the ontological and epistemological implications for statistical analysis is necessary when moving from the space to the place domain. In this context, he raises the question of what kinds of statistical analyses are even feasible in this case. This, the assertion that “[g]eometry is not geography” (Bergmann and O’Sullivan, 2024: 49), and recent proposals on place-based GIS (Gao, 2022) and platial information science (Mocnik, 2022) form the starting point for the following discussion.

The remainder of this article discusses conceptual matters pertaining to the alignment of place experience with statistical inquiry. The introductory quote from Alice in Wonderland illustrates a central challenge, namely, how meaningful experiences can be represented in a form compatible with statistics. In order to address this challenge, the following two sections, “The humanistic notion of place” and “Conceptual commitments,” present the humanistic-geographical concept of place, explications of some of its central tenets, and a wire rope metaphor. Following these conceptual foundations, the section “Identifying random components” discusses the extent to which the introduced concept of place is generally subject to randomness and can therefore be considered for statistical investigations. An attitude we should take in doing so is suggested in the section “Attitude toward the mode of inquiry,” before stochastic processes and thus potential statistical objects of investigation are presented in the section “Stochastic processes.” My overarching goal with this article is to create a conceptual basis for the further development of specific methodological approaches to the statistical study of place.

The humanistic notion of place

Until the 1960s, the concept of place was largely used in geography in a descriptive, idiographic sense. This was mainly due to the regional-geographic approaches that had prevailed until then, but since the last 40 years of the 20th century these have been supplemented by more general concepts of place (Cresswell, 2015: 30 ff). One of those is the place notion from the humanistic tradition, which emphasizes meanings, values, and agency, and is based on principles from existentialism and the attitude called phenomenology (Seamon and Lundberg, 2017; Tuan, 2006). Phenomenology focuses on the experience of everyday, taken-for-granted phenomena that make up our so-called “lifeworlds” (Zahavi, 2003: 125 ff). ² This perspective is grounded in the assumption that people are naïve realists when experiencing the immediate world around them: “[…] we take the existence of the world outside of us for granted. In daily life we do not primarily encounter impressions or feelings; things beyond us are the dominating factor: the places we go, the people we interact with—family members, side-walks, bicycles, computers, coffee cups” (Yoshimi, 2016: 5). The focus of phenomenology is thus on pre-reflective experiences understood as subject–object relations, whereby the objects—in the context of this article—are given as places in the form of psychological phenomena that are holistic entities inseparable from the world “out there” (e.g., through the sensory experience of it), since we are always-already emplaced (Seamon and Larsen, 2020). Phenomenological inquiry is based on the premise that the only valid access we have to any relationship between subjects and objects as outlined is to explore the latter without prejudice and free of any assumptions. Only the pure experience of places from the perspective of the experiencing subject is valid in order to eventually reveal the true essence of the world as it is experienced (Kockelmans, 1994, pp. 89 ff).

The essentialist term essence that I use above is key as it points to the social-ontic ³ status of places in humanistic works: following from the phenomenological goal “to arrive at a consciousness of pure essence” (Buttimer, 1976: 279), places are seen as fundamentally characterized by essences, the existence of which is considered necessary to give a place its character. These essences are supposed to exist immutably (though we will relax this attribute in the following section) and can thus be “discovered” through “living” places (though discovery does not need to happen in a direct manner), which gives essentialism a realist stance. The focus of inquiry in humanistic geography on experiential worlds implies that these essences are not seen as transcendent, metaphysical properties of the world “out there,” but as qualities of the concrete bond between world and self. ⁴ The assumption of the existence of essences gives rise to another fundamental property of humanistic-geographical place research: the assumption of the existence of a non-contingent and invariant structure in the pre-reflective, unquestioned geographical dimensions of human experience (Seamon, 2018: 12–14). These structures refer to the lifeworlds briefly introduced above, which function as a basic structure upon which human life unfolds, including experiences of place. The aim of humanistic research is to grasp this basic structure in a holistic sense, that is, without considering particular essences in isolation (as is done with psychometric approaches), but finding out how they are together, can be together, and give rise to attributions of meaning as well as the contingent characteristics of place experience.

Applying the phenomenological framework of inquiry and an essentialist ontology, a number of core elements, the experience of which jointly characterizes places and their associated lifeworlds, have been identified:

• location (the “where” of a place that contributes to its context; Agnew, 1987),

• locale (the material settings of places and for social relations; Agnew, 1987),

• time (place can form immediate or over some duration of time; Raymond et al., 2017),

• scale (possibly ranging from a corner of a room to the globe; Tuan, 1977),

• reiterated mundane practices that are partly constitutive of a place (including bodily movement, which has been referred to as the “place ballet”; Seamon, 1980),

• sense of place, understood either as emotional attachment (Agnew, 1987) or as literal sense-making of a lived geographical environment (Seamon, 2018),

• topophilia (a strong affective bond between self and place; Tuan, 1977),

• sense of community (the feeling of a place facilitating social incorporation; Relph, 1976),

• sense of belonging/“rootedness” (the capacity of a place to satisfy the psychological need to belong somewhere; Cresswell, 2015; Seamon, 1980; Relph, 1976),

• gathering (of feelings, things, memories, etc.; Casey, 1996),

• landscape (understood as the visual impression of an immediate, material geographical configuration in which the observer is not directly involved; Cresswell, 2015; Merriman et al., 2008),

• memories (which can be evoked in and by places, are potential amplifiers of sense of place, and can be used to sustain and promote certain senses of place, for example, when creating collective places of public remembrance such as the 9/11 Memorial in New York City or the East Side Gallery in Berlin; Till, 2005; Casey, 1987).

An element that affects place without being a direct element of it is ambient space into which places are embedded (Tuan, 1977). Ambient space is here interpreted as the undifferentiated fabric between the locations of places serving to connect them through movement. This argument shows that no simple dichotomy between space and place ⁵ exists and that both concepts more often than not appear intertwined in various ways.

Conceptual commitments

Humanistic geography has been significantly influenced by phenomenology. The core idea of phenomenology is to ignore (but not necessarily dismiss) a supposedly “real” world behind our human conception and experience of it and instead strive to understand the everyday, taken-for-granted world as it reveals itself to us (for recent comprehensive overviews, see Gallagher, 2022; Zahavi, 2019). Zahavi (2019: 44) describes the aim of phenomenology as “to investigate the essential structures characterizing our experiences, their correlates, and the connection between the two.” Place is hence placed in experience including consciousness and intentionality (which, here, means experience-of and consciousness-of). The way in which such inquiry can be achieved varies between important proponents of phenomenology such as Edmund Husserl, Martin Heidegger, Jean-Paul Sartre, and Maurice Merleau-Ponty, and I refer the reader to the article by Jacobs (2018) for a detailed comparison. The aim of the following sections is to lay the groundwork for the development of self-contained statistical perspectives that are compatible with the humanistic conceptualization of place. This section therefore focuses on two aspects of phenomenology relevant to the present discussion that transcend its various construals: an essentialist stance, and wholeness and holism. The reason for this selection is that many other features of place are derived from them: essentialism provides the metaphysical basis for what characterizes place in the humanistic school; and the neglect of the wholeness of place is the most frequently criticized trait in positivist approaches. Statistical approaches should therefore respond appropriately to these aspects in order to (a) avoid operating with an inherently incoherent or eclectic conceptualization of place, and (b) ultimately avoid facing the same criticisms that have discredited many previous formal and quantitative approaches.

Essentialism

In its most generic form, essentialism denotes that categories differentiating objects are based on sets of underlying natural (not just fabricated) properties called essences, which are considered real (beyond the purely phenomenal) and responsible for the identity of the respective categorized objects (Cartwright, 1968; Gelman, 2003). The kind of essentialism ⁶ that seems a natural candidate for the context of humanistic place and statistics is psychological essentialism. As pointed out in the Section “Humanistic notion of place,” the humanistic tradition is concerned with place as a psychological (but not exclusively mental) phenomenon. The corresponding psychological essentialism holds that essences are found not “somewhere in the world” but in people’s mental images of the world (Gelman, 2003; Neufeld, 2022). This makes psychological essentialism a useful assumption for characterizing so-called social kinds ⁷ such as places, as these, following Lefebvre, are produced (by society, individuals, etc.) instead of created like natural kinds ⁸ (Bross, 2021). The distinction put forward between metaphysical essences (which would be properties “out there”) and those in mental representations (which are properties of representations of the world) is an important one. As Richardson (2011: 3) notes, “[e]ssentialist ideas are more likely to linger in popular culture rather than in […] scientific practice.” For example, although there is no scientific evidence to support the assumption of the existence of human races, some people may nevertheless essentialize the concept of race (and factually incorrectly in relation to the world “out there”) in their mental representations, leading to the cultural phenomenon of racism (see Zack, 2000, for a discussion of this issue with reference to essentialism). This example drastically illustrates why the aforementioned distinction, and thus the main idea of psychological essentialism, is crucial. Although psychological essentialism is not uncontroversial for a variety of reasons (see Berg-Sørensen et al., 2010; Kronfeldner, 2021; Zachar, 2022), it is based on strong evidence that people do essentialize many everyday categories in the form of concepts, ⁹ an ability that seems to develop in childhood and then continues into adulthood (Mandalaywala, 2020; Neufeld, 2022; Rehder, 2017; Rhodes and Moty, 2020). In doing so, people may not always know the precise nature of essences (causing so-called placeholder essences to occur; see Strevens, 2000), but they often presuppose the existence of properties that give places and other social and cultural constructs their identity.

Newman and Knobe (2019) group psychological essences into two broad categories: causal (or Aristotelian) and ideal (or Platonic). Causal essences are abstract qualities that may not be immediately observable but that are considered to cause the perceivable features of objects of a particular category. The concrete but non-essential features are thus contingent and, when detached from the causal essences, are not sufficient to decide membership in a category. Causal essences thus evoke the structure of a category in a cause–effect manner. In contrast, ideal essences are underlying values or ideals embodied by the visible ephemeral features. The latter are therefore regarded as instantiations that actualize the underlying values or ideals that give members of a category their identity. The perceivable qualities are hence not caused by underlying abstract essences, but they instead reify the latter in the sense of a reproduction. A third type of essence, originally introduced by Medin and Ortony (1989) and described in concise form by Strevens (2000), is that of statistical essences. Thereby, some essences may only be statistically related to observable properties but may not suffice for actual category membership. For example, Strevens (2000) refers to the widely held believe that birds fly, which appears to be statistically true in the everyday perception of many people despite there being birds that do not fly. Thus, category membership under this relaxed understanding no longer depends entirely on necessary essences that give rise to observable properties. Instead, some statistical essences may simply correlate with contingent properties observed by those who essentialize.

For the remainder of this article, I argue that it is reasonable to think of people as naïve ¹⁰ essentializers in their everyday lives and thus also when it comes to places understood as partly mental entities. Therefore, it seems useful to adopt the notions of placeholder and statistical essences. In everyday life, people gather cues from their immediate (but not necessarily only local) surroundings. They weave a web of meaning, affection or aversion, action potential, etc. from these cues and in interaction with memories, cultural imprints, and so on. According to the so-called inherence heuristic (Cimpian and Salomon, 2014), people tend to ascribe inherent properties to what they experience and look for matching essences in what is directly accessible to them. However, this often remains limited to a general assumption of essences without concretizing them (leading to placeholder essences). The attempt to concretize also often focuses on recurring direct experiences and makes people consider potentially contingent but frequently occurring typical characteristics as essential (which gives rise to statistical essences). These viewpoints lead, in the words of Newman and Knobe (2019: 600), to in-the-moment explanations, which, if experienced repeatedly, may lead to a mental image of place. The basic principle of these in-the-moment explanations, that is, making decisions in an immediate context and directly influenced by it, is consistent with the basic tenet of behavioral geography that the environment in which people live has an influence on their behavior, even if this does not always lead to the most rational decisions, as can be the case with statistical essences, which may be objectively flawed (see Golledge, 2008, for an overview of behavioral-geographic contributions). ¹¹ The proposals put forward so far are also in line with Doreen Massey’s progressive sense of place, which is a critique of the often rather static, more traditional approaches of humanistic geography (Massey, 1991, 2005, 2008). Statistical, psychological essentialization implies that people have the possibility to change the underlying process (since essences are not just somewhere “out there” hence inaccessible and immutable), and place-making based on naïve in-the-moment explanations and the assumption of statistical essences also allow for flux.

Wholeness and holism

A main critique of humanistic geographers about objectification and quantification as it happens in psychometrics and related fields is not the representation of people in terms of numbers or regularities, but the reductionist way in which this is done (Seamon, 2018). As we have seen in the previous subsection, the focus of humanistic inquiry is on human-being-in-the-world and thus on an inseparable whole. The property of wholeness is thus paramount and cannot be dispensed in any serious attempt to learn about the complex entanglement between individual, environment, attachment, agency, and meaning unfolding upon experiences of lifeworlds. But what does wholeness mean? I argue that there are two forms of wholeness (or “irreducibilities”) in relation to place. David Seamon (2018: 21) argues that place is neither objective nor subjective. Instead, he suggests, place is given as a lived engagement that is both shaped by the world and shapes the world. This describes a specific kind of wholeness: Rather than decomposing place into subjective and objective components, it is proposed to address the structures of the lifeworld that are interwoven with the ways in which people engage themselves in experiencing and living both internally and externally. It is appropriate, therefore, to assume a structure that reaches both into the interior of subjects (subjective) and, simultaneously, into the objective externalities of experience, enabling that structure to serve as the setting up of a possibility space for engaging with the world and one’s own placement in it. This way of putting it implies that place reaches out into both the inner subjective condition of humans and into lived contexts. The latter means that place is always-already in a relation with context, and a change in lived context thus implies a change in place (Slife, 2004). This inseparable intertwining forms one kind of wholeness.

A second kind of irreducibility relevant to place arises from the notion of holism and concerns the way we describe and know place. Most psychometric analysis provides us with piecemeal information and does not directly reveal anything about place as such. Instead, it only allows us to know about the components of place in isolation or to search for causal explanations horizontally, at one analytical level. MacKinnon (2022) refers to such approaches as downward and horizontal reductionism. ¹² The counterpart to such positions is holism, which postulates that knowledge of the whole is more than the knowledge of the sum of the constituent parts. This attitude is not exclusive to the humanistic school, but can also be found in postmodern approaches such as actor-network theory and assemblage theory (Müller, 2015), where it is even more pronounced through notions such as emergence (Zahle and Kaidesoja, 2019) and exteriority of relations (DeLanda, 2006). Similarly, complexity theories, which generally have much in common with the postmodern approaches referred to above (Spies and Alff, 2020) also encompass holism, and in particular the version that Steven Manson labels “aggregate complexity” (Manson, 2001; O’Sullivan, 2004) bears similarities to the understanding used here. The concepts of holism that are relevant for the present discussion are epistemological and methodological holism (Ralston, 2015). Both mean that we should not look at the individual components if we want to learn something about a whole, whereby epistemological holism concerns the way we acquire our knowledge about a whole and methodological holism concerns the more practical aspects of inquiry. Based on the discussion in the previous paragraph, this point of view is a logical extension of the concept of wholeness described above, which concerns the nature of place and as such implies an ontological holism. If a place is a whole with a meaning of its own, then it follows that ideally we should not dissect it. In other words, wholeness, as introduced in this section and adopted from Seamon (2018), is a statement about what a place is, and holism is a statement on how we can learn about it.

Intermediate summary

We shall briefly recapitulate the various aspects discussed above and synthesize them into a coherent picture, specific enough to proceed with our main discussion. In the humanistic school of thought, place emerges from ongoing and recurring experiences of lived engagements, and is nourished and consolidated by them. These experiences take place through intentional consciousness, which is directed towards sensually perceivable and discernible units of experience. Even though intentionality may be directed towards something very specific, the experience as such is always shaped holistically by the totality of lifeworlds, situations, memories, and other elements, even though some elements may sometimes remain in the background. Following Seamon (2018) and the above discussion of phenomenology, it makes sense to consider humanistic place as a link between the inner, subjective and outer, objective domains of life. Figure 1 illustrates this idea. The exact shape of this bond linking the subjective and the objective is flexible and subject to internal and external forces and their interplay. Figure 1 shows this with dashed lines that point to other possible pathways of experiential bonding. It seems useful to think of place (the entire constellation shown in Figure 1) as a dormant, but potentially latently effective structure, which is the result of previous lived experiences and is anchored in our mental image of the world. This latent, dormant structure forms a horizon of understanding for new lived experiences with the relevant constellation of elements (and possibly new ones), which then leads to an enactment of this up to then dormant structure, with potential changes possible. This possibility for change does justice to more recent ideas of humanistic place, inspired by Doreen Massey’s work (Massey, 1991, 2005, 2008). The character and identity of such places as they reveal themselves is defined by the statistical form of psychological essences, assuming that humans are naïve essentializers who, based on concrete experience, ascribe both clear, nameable and opaque, hard-to-name qualities to their mental structures that qualify as place. The decision to use statistical essences also implies a mesogeographic stance on place-making (see Miller, 2018), as it assumes regularities in relation to individuals or groups, but does not lay claim to universal laws. This is an important feature of the conceptualization developed here, as the latter has often been criticized at spatial science.

Figure 1.

Illustration of the place concept as used in humanistic geography; all elements mentioned for the subjective and objective domains are indicative; the dashed gray lines indicate examples of different possible bond shapes.

One metaphorical way of looking at place as stipulated above is to think of it as a wire rope made up of strands, which are themselves made up of individual wires in return. In this picture, the experiential bond in Figure 1 is the stylized rope, which would indeed appear as intertwined components, similar to the rope shown in Figure 2(a). Like places, wire ropes have properties that are not readily present in their component parts. For example, resistance to tensile forces only develops through the interaction of the components of the wire rope as a whole. Similarly, a place can be strong and cohesive (e.g., with long-standing attachments) or weak and loose (if there have been only a few or superficial engagements). Furthermore, a wire rope is only as strong as its two ends are attached on the connected sides. A firm rooting in the inner and outer realms of place is also needed for the latter to be strong, and too much alteration of the mounting on one side can lead to a weakening of the bond. It may also happen that a wire rope becomes unraveled, as shown in Figure 2(b). This can be an opportunity to rewire the rope in a slightly different, possibly stronger configuration. We can think of the act of place-making in terms of direct experiences as such a situation: an existing place, already lived and re-lived a few times, is changed by new experiences that leave the bond in a slightly different state than it was before. However, looking more closely at Figure 2(b), some wires remain bundled, making sure that the rope does not snap. These could be our essences, which remain in place and only change after a long time or with drastic changes in the sense of statistical essences.

Figure 2.

Two examples of wire ropes. Place, as conceived in this article, can be compared metaphorically to the cohesive force and possible rewiring of wire ropes connecting two spheres at their respective ends. (a) Cohesive wire rope; and (b) unbundled wire rope.

The above summary and metaphorical image imply two separate but closely related expressions of place. One is the concrete, direct experience of place, which encompasses cognitive, embodied, material, and other aspects. The other is a mental image of place, which can be seen as a reduced and tightened bundle of meaningful lived experiences and is more than just a collection of memories. The first expression constitutes the object of inquiry in phenomenological studies. The second expression is the leftover of place that we carry in our mind and that predisposes our next concrete lived experience of a place. The advantage of conceptualizing place this way is that we avoid strong ontic claims about the nature of place being a “thing” out there and thus any metaphysical stipulations that may open numerous doors for (justified) critique. The target object at the center of statistical inquiry therefore is the structure that forms, and is formed by lived, meaningful geographical engagements. Having clarified some key conceptual underpinnings, the following section discusses where randomness can be found in the outlined place concept.

Identifying random components

The development of a statistical perspective on the humanistic notion of place only makes sense in the light of the presence of randomness because statistics “relates […] empirical facts […] codified and structured into data sets [with] hypotheses […] formulated in terms of probability distributions […]” (Romeijn, 2022). This section discusses where randomness can be found in a humanistic place context. ¹³ It is indeed impossible to list all conceivable specific random components of place as these would vary with each individual case. In this section, I thus focus on general ways in which randomness can be found in humanistic place as sketched out above. Furthermore, the following two paragraphs motivate the types of stochastic processes proposed in a separate section “Stochastic Processes” (and in the same order as the two paragraphs here).

When we experience place, our consciousness can wander between different experiential focal points of the place (which follows from conscious intentionality, see above), but the set of all other possible focal points is always co-present and affects the experience. This process of wandering in focus is most likely inherently stochastic. For example, when we feel at home, we may focus on our refrigerator, whereas when we engage with the same place again, we may turn to our desk or to something intangible, such as a particular thought. Seldom, though, will our focus and experience of place (and thus place itself) be identical, and because of its advocacy of pre-reflective, immediate experience, the humanistic-geographic literature, with its focus on phenomenology, does not suggest a deterministic mechanism either. Further, in all cases described above, our foci are also not independent of all the other things, smells, feelings, etc., which are always implicitly operating and in which the lived conscious intentionality is embedded, but which are pushed into the background due to some momentary focus. These background elements are located at both ends of the experiential bond in Figure 1 and the way they are effective in a concrete experience of place is co-constitutive with the realization of a particular experiential bond. This shows that a one-sided statistical focus on only the various focal points that one can have (e.g., by counting and deriving a histogram) or on the influence of selected quantifiable background or contextual elements (e.g., using a regression analysis) does not reveal the latter simultaneously acting interconnectedness that is so important for characterizing place experience. The randomness inherent in a place as such can therefore only be observed by considering entire constellations (i.e., the entire illustration in Figure 1) as indivisible units and how these are stochastically reconfigured by conscious intentionality during their experience. The main argument I am putting forward in this paragraph is to characterize “place in action,” that is, its momentarily experienced form, as being largely stochastic. Expressed in the words of the wire rope metaphor, the outlined stochasticity lies in the rewiring capacity of the loose wires in Figure 2(b) and their interplay.

The wandering focus and embedding mentioned above concentrate on randomness in the concrete living of places. A complementary perspective on place and randomness is the focus on essences. We have established that for the present case, these are of a psychological and statistical nature. One consequence of this conceptual decision is that the actual key character of what makes a place a certain place is itself stochastic in nature. If people are considered naïve essentialists, forming essences based on what they experience recurrently, there is no straightforward determinism at play, so the essences formed are most likely outcomes of a stochastic process, which may be conditioned by certain circumstances such as cultural, socioeconomic, and demographic characteristics. This stochastic nature is a major difference to the (in principle also possible) assumption that places can be characterized by more metaphysical forms of essentialism, which would force us to believe in the existence of immutable essences somewhere “out there,” and thus in a preconceived and deterministic nature of these essences that can only be revealed (but not changed) by living places. A joint perspective on stochastic essences and the already outlined stochastic bonding between subjective and objective aspects of place and its mechanics may enable us to examine which elements this bonding links and why, how they are linked in a particular case, how the interplay of the elements linked by the bonding actually unfolds, and ultimately to draw conclusions about possible essences. All these questions can be answered probabilistically, since, due to our chosen conceptualization, they are not the result of deterministic processes.

Attitude toward the mode of inquiry

My aim in the following two sections is to provide perspectives on what I consider basic for preparing the ground for an operational statistical perspective on place. I will begin with a general statistical attitude toward place in the present section before moving on to the question of what we might focus on and how, yielding outlines of candidates for adequate stochastic processes. My aim is not to provide a concrete cookbook but to sketch out a framework allowing us to eventually arrive at a meaningful statistical treatment of place.

A major reason why quantitative and formal inquiry including statistics has been largely rejected by place scholars seems to be the nature of the operationalization associated with it. Operationalization establishes a link between a theoretical concept (here: humanistic place) and an empirical concept, that is, a concept that, when put into practice, allows for the derivation of observable variables (Roskam, 1989; Sargent Weaver, 2015). Operationalizations associated with spatial statistics and GIS often rely on the assumption that discrete objects and continuous fields can be reduced to so-called geo-atoms (Goodchild et al., 2007). This entails not only reductionist forms of representation (e.g., point, line, polygon), but also reductionist methods that decompose objects into individual components for their isolated investigation. The latter is usually referred to as analysis, which originates from Greek and literally means to dissect something in order to understand it better. Gahegan et al. (2001) argue that the choice of a particular operationalization (in a GIS context) is usually a compromise between the specific needs of a researcher and the type of data that can realistically be obtained. The circumstances, objectives, or, in short, the general modus operandi of our research thus compel us to adopt an analytical approach.

We shall remind ourselves of the discussion of phenomenology and the nature of place discussed above. The way we have established place conceptually is as a link between inner and outer experiences of geographical engagements, giving rise to a lived nexus of meaning. Kant, in his “Critique of Pure Reason,” remarks that something that has been decomposed was once together and not only somewhere out there in some inaccessible, assumed “reality” behind phenomena, but right there, in our experience (Kant, 1787: 130). Otherwise, he notes, what we have decomposed would not have had the meaning that we want to learn about through the decomposition. In the case of place, this is rather evident. If place is a web of meaning and experience identified by essences, then the mere presence of the individual components, without being embedded in and forming a meaningful whole, would hardly have prompted us to conduct a dissecting analysis. What we can learn from this Kant-inspired consideration is that even analysts who take a methodologically dissecting approach have first recognized an overall, synthetic whole. Consequently, and in analogy to the reduction or bracketing in phenomenology, to address this wholeness, we need to consciously embrace this initial holistic step, resisting the reflex to dissect. Instead of postulating novel forms of platial analysis ¹⁴ (in analogy to spatial analysis) I therefore propose a different form of investigation that may be called platial synergetics. This proposal intellectually follows David Seamon’s distinction between analytic and synergistic relationality (Seamon, 2018). Adopting the latter, the emphasis is shifted terminologically from the piecemeal dissection of place to a clear focus on synthesis, and thus on maintaining wholeness, without losing the conceptual connection to humanistic geography. In this sense, Seamon (2024) argues that analytic relationality aims to impose external connections on parts that already belong together—which, by and large, describes the business of psychometry—while synergetistic relationality assumes the co-constitutivity and indivisibility of parts and whole.

One implication that follows from synergistic relationality and the notion of the co-constitutivity of parts and whole is that parts that are assessed in isolation are not the same as parts-in-place. In psychometrics, place is often operationalized through surveys that assess aspects such as place attachment, place identity, and place perception. Although these aspects may be interesting in themselves (and find their justification in psychological research, for example), they are somewhat different from what they yield in interaction. For instance, it is often difficult to separate place attachment, that is, the way people become emotionally attached to places, from place identity, since the two probably do not develop separately from each other but are to a certain extent co-constitutive. What we are doing in analytical approaches, I argue, is assessing and analyzing constructs that have a different meaning than we might intend to grasp. Furthermore, as David Seamon points out, these near-relevant, yet slightly amiss parts in isolation are re-synthesized together in regression modeling, structural equation modeling, or other mechanistic forms. In other words, we start with a whole that we want to learn about, evaluate slightly aberrant parts, and reconfigure them in ways that may be somewhat artificial, if not outright guessed.

Another implication of synergistic relationality and the phenomenological concept of place is that we cannot find what we are looking for in isolated parts, because places do not reveal themselves to us in this way. Recall that, for the scope of this article, we are interested in place understood as a long-term structural bundle of recurring experiences of geographical engagements as they reveal themselves to people. However, we do not usually have partial experiences of places, so neither direct encounters with phenomena nor the resulting essentialist structures can be found in isolated parts. We should consider place as a whole revealed and solidified by recurring phenomena, as a unified entity that cannot be decomposed. A helpful analogy for this can be found in spatial statistics: we would not normally break down a census polygon into its enclosed points because the zone as such, including the variables attached to it, has its own meaning. For example, average income mapped at census tract level cannot be understood by looking at individual coordinates, and any attempt to do so would amount to an ecological fallacy (see Walker, 2021). In this sense, and with regard to place, we should look for randomness and structure in the overarching whole and be careful not to tear it apart.

The attitude that I propose for platial synergetics could be described as upward synergistic abstraction. If places, as discussed above, do not reveal themselves to us piecemeal and if isolated parts do not contain what we are looking for, what we do when we look at places in their parts could be described as an analytical abstraction downwards, that is, simplification and making tangible in a dissecting way. In statistics and other formal, quantitative approaches, this basic attitude often goes unscrutinized and is instead accepted as a pragmatic approach to simplifying a problem. This attitude may be appropriate when the object of study presents itself to us in its full complexity. In the case of place, however, the object of study is not readily revealed to us as researchers, but merely as a collection of individual parts. The meaningful complexity of interest remains hidden from us at first. If we start simplifying the piecework by breaking down the problem, we are only abstracting the individual parts but not place as such. Instead, we would have to take the opposite direction and abstract upwards, since only the superordinate structure of place enables us to order the clutter of individual parts in a meaningful manner. I therefore suggest reversing the traditional analytical mindset and taking a synthesizing approach that seeks upward simplification. This may sound like upward reductionism as introduced by MacKinnon (2022) and discussed above. However, I do not mean by upward synergistic abstraction that we reduce. Rather, my aim is to move attention from the inadequate level of pieces to the adequate level of their ordering and combining whole, which would result in simplification such that it would, in fact, enable us to find what we are looking for in the first place. What I do also not mean is to conduct a full-blown phenomenological study in preparation for a statistical investigation. Rather, my point is to focus the researcher’s mind on a particular mode of operation at which to proceed with further inquiry, much as the analytical mind focuses on dissection and reassembly once that mindset has been adopted.

Stochastic processes

Statistical methods establish a link between observations and models of structured indeterminacy, expressed in terms of probabilities (Romeijn, 2022). Operationalizing this requires us to think about reasonable hypotheses that we can posit and how we can assess them in the light of suitable data. However, for any of this more practical activity to be meaningful, we need place-related stochastic processes that formally describe the object of study about which we are hypothesizing. In line with the preceding discussion, I propose two types of stochastic processes: one (i) that formalizes the momentary experience of a place and another (ii) that captures the bundling of a place into a dormant mental construct. In general, a spatial stochastic process as used in spatial analysis is defined as a collection of random variables, $\gamma =\left\{X_s:s\in S\subset {\mathbb{R}}^d,X\in \mathrm{\Omega }\right\}$ (Westerholt, 2022: 40), where the X are random variables that are assigned to elements of a geometric index set S representing units such as points, lines, or polygons. Further specialization into geostatistical, lattice, and point processes then depends on the properties of the index set S (Cressie, 1993: 8 ff) while the sampling space Ω is deliberately kept flexible to accommodate a wide range of possible random variables. The following sub-sections discuss the two processes introduced above by characterizing their associated index sets, not in a technical sense but in conceptual terms.

Stochastic processes for place engagements

Stochastic processes of type (i) concern the act of experiencing place. Thereby, we consider the re-enactment of the dormant structure already discussed and illustrated above as a wire rope as a remnant of previous place experiences. This structure serves the concrete experience in question as a horizon of understanding, since it predisposes the new experience to a certain extent. Drawing on our conception of place as introduced in Section “Conceptual commitments,” including ideas such as in-the-moment explanations, the inherence heuristic, and the concept of statistical essences, this structure is subject to potential change in every concrete re-enactment and experience. In this process, tightened structures must loosen up to a certain extent and interact, similar to the unraveling of wire rope as illustrated in Figure 2(b). This interaction not only takes place between the component structures of the tightened bundle itself, but also involves the fabric of some lived experience beyond the already established and re-enacted place bundle, since neither bundled past experiences of places nor the mere here-and-now are sufficient to understand lived place as it reveals itself to us. What is at issue here is not the creation of a new hardwiring (that would be type-(ii) processes) but rather the re-enactment and invigoration of place on the basis of an existing bundle upon which new place experiences unfold.

Given the wholeness property and following the upward synergistic abstraction introduced above, we should direct our interest to the most expansive possible level of abstraction. Rather than focusing on isolated place features such as the color of a pavement, a person’s five strongest local social relationships, or the strength of a person’s preference for a tree—as we would in psychometrics—we shall refer to units that represent many such aspects in combination rather than individually. For example, the three properties mentioned above may interact and together generate what ultimately constitutes a place for someone. Overcoming the dissection and working with a bundle without giving too much emphasis on its constituent parts in isolation right from the outset is what is meant by synergistic upward abstraction. Which concrete bundles are to be formed in an informed or exploratory way ultimately depends on the particular investigation. The key to traditional spatial statistics, meanwhile, is to define what are known as spatial weights that establish links between the elements of the index set (see Getis, 2009). For processes of type (i), I propose considering three types of such links between bundles: links from the tightened place bundle to the products of the momentary living of the place, reciprocal links back to the bundle, and self-interactions. The first category of links describes how strongly (subsets of) previous and firmly consolidated experiences of a place affect the current circumstances of a re-enactment of a place. The second category of links indicates how the current experience of a place may contribute to untighten the pre-existing bundle from earlier encounters with the place in the sense of the wire rope metaphor. The third category of links represents effects that can be expected either only from the pre-existing bundle or from the current experience of the place. The assumption of this outlined model is that place experience as such is composed of the way in which our established understanding of a place is active, how it can be eroded or changed, at least for the moment, and of the self-preservation forces of the established place as well as of the very moment itself. In a technical, formal sense, this proposal would yield a non-symmetric matrix of weights connecting bundles of place characteristics and with inertia on the diagonal. The simultaneous presence of weights connecting a re-enacted place bundle and the here-and-now stresses the co-constitutivity of concrete place experience, shaped by our predispositions towards a place and what presents itself to us right before our eyes. Such formal weights could be adjusted to test relationships of how living a place works in more detail and at the holistic level rather than at the level of its atomistic constituents.

Stochastic processes for dormant place bundles

While processes of type (i) are based on the partial unbundling and subsequent reconfiguration of what is left from previous encounters with a place, the focus here is on the dynamic constitution in what I have considered above as units: the bundles of wires and how these are reduced and tightened up into a mental image that remains and can be re-enacted in future place encounters. Because, according to our outlined place concept, statistical essences are formed through recurring experience, they must remain in the dormant structure in the form of intermediate steps of their evolution. Moreover, since humanistic geography is concerned with asking how essences can be together (Seamon, 2018), they cannot just merely be together, but seem to be subject to structuration. Type (ii) processes refer to the generative mechanisms that produce such constellations over many steps of reliving a place. Figuratively speaking, we are looking for processes that generate the profile typologies of wire ropes (see Figure 3), which does not happen in a single step but over many iterations. In order to operationalize processes of type (ii), it would thus not be enough to observe a single outcome but rather a chain of what remains from multiple place experiences. While processes of type (i) assume a given pre-structuring of place (i.e., a concrete realization of an intermediate step of the evolving dormant place structure), processes of type (ii), instead, concern the reordering of the dormant structure itself. Expressed in analogy to spatial statistics, processes of type (ii) are thus akin to point processes as encountered in stochastic geometry, but without any geometry in mind and focusing on the idea of the units as such being stochastic in nature. What interests us here is a relative configuration of bundles that persists beyond the here-and-now of living a place. I propose to consider the way essences are and can be together via the notion of homogeneity and by analogy with so-called empty space and nearest-neighbor statistics as used in point process theory.

Figure 3.

Illustration of two dormant place bundles, inspired by typologies of wire rope cross-sections as presented in da Silva (2022). The red hexagons visualize essences, while the white circles around them represent contingent properties. (a) Structurally weak; and (b) structurally solidified.

In point process theory, there are two common approaches for modeling processes with spatial structuration: cluster and Gibbs processes (see Illian et al., 2008: 371 ff). The first option models points without interaction, but with a source of clustering, for example, seedlings (a so-called daughter process) falling around a tree (whose location would result from an associated parent process). The second option models proactive interaction between points, as may be appropriate when modeling territorial behavior in animals. What we want to statistically model is how the essences are arranged and interact to produce the holistic nature of a place, and in addition how some selected contingent features contribute to this and are arranged in relation to the essences. I argue that, following the humanistic notion of place, we would need to think of processes of type (ii) as containing a Gibbs-like parent process with some independent daughter processes imposed. Essences giving rise to the attribution of meaning is a strong indication that the essences do proactively interact. This, and considering the essences as central components on which the contingent criteria depend, is why I suggest that we consider a proactive interaction component in our stochastic process for type (ii). In contrast, while in the concrete living of a place (see the notion of weights introduced for type (i) processes above) the contingent properties are likely to interact both with each other and with the essences, I suggest that we consider their role in the remaining bundle to be less proactive but still arranged in relation to the essences (since, as stated above, the latter lead to the former). Contingent properties may thus be regarded as daughter processes, analogous to the seedlings in the example above, arranged around bundles of essences.

My suggestion for a statistical understanding of how essences are in terms of their efficacy in maintaining the identity of the whole is the concept of homogeneity. Two lines of thought are possible: the assumption that essences (understood as bundles of properties) are particularly efficacious in a homogeneous form, where they consist of only a few components; or that their efficacy is particularly strong precisely because of the heterogeneity of a large number of interacting components, resulting in greater resilience to structural changes in the statistical essences. Deciding which of these two interpretations actually seems more appropriate would be a question that we could support by means of a statistical consideration of homogeneity. In the case of the proposed Gibbs-like modeling assuming proactive interaction of essences, the homogeneity or heterogeneity would be tied to so-called pair potential functions, which, in point process theory, indicate how the presence of a point acts attractively or repulsively on the presence of other points in the vicinity (Dereudre, 2019: 185 f.). Such functions could reasonably be imagined for formalizing the potential of mutual attraction of essences. For instance, the essences shown in Figure 3(a), which represent a weak place bundle, could suggest pair potential functions that repel or only weakly attract, while the more stable configuration in Figure 3(b) could indicate a strong pairwise attraction between the essences. Pair potential could then be used as a formal representation of the assumptions about how essences can mutually facilitate each other, come together, and, as expressed above, be together. Such potential functions would, albeit in a different form and with a slightly different goal, take on the role of weights as introduced for processes of type (i). Exploratively, the estimation of models of this kind could allow the identification of possible pairwise potentials (and, via successive bundling, also between more than two essences). Confirmatory research could test informed specifications of potential functions against collected evidence. The proposed approach to addressing the question of how essences can be together is to adapt potential functions as used in Gibbs processes towards bundles of essences.

Another way of characterizing the way essences can be together is to consider so-called empty space and nearest-neighbor statistics (van Lieshout, 2010: 265 ff). Both place as a whole and the essences are considered in humanistic geography as a result of the recurring living of places. The evolving essences could thus be imagined as a web of initially loose structures of properties, some of which become increasingly bundled up through recurring experiences, thereby forming persistent bundles that we may eventually call essences. The cohesive forces described in this way can be approached figuratively either through the empty spaces between essences or through the dynamics of other wires coming closer. Figure 3(a) illustrates a composition in which more empty space is likely to be present between both the essences and the contingent properties than would be expected under random conditions and under the assumption that this is an actual place bundle (and not just an arbitrary spatial experience, which may resemble a tangle). In contrast, Figure 3(b) may yield a statistically significant nearest-neighbor statistic. It would be useful to establish how such empty spaces or proximity measures should be adopted from the geometric metaphor and designed for humanistic-geographical place as considered here. Ideally, such a measure would be in correspondence with the pair potential functions mentioned above, which would also have to include some form of proximity assessment. The goal would then be to be able to assess the emergence of certain properties and their interrelations over several concrete experiences, using statistics about their interstitial spaces and neighborhood relations. In this way, we could, on the one hand, statistically approach candidates for essences. On the other hand, we could also observe essences that have already been identified (or assumed on well-founded grounds) over the lifetime of a place, in order to ultimately be able to draw conclusions about the dynamics of the latter.

The above remarks on processes of type (ii) appear to be somewhat analytically contrived and thus seem to contradict the idea of upward synergistic abstraction. However, what is at issue here is the diagnosis of the “machine room” of the holistic overall domain of place. We are not trying to break down individual components of place but are still operating with wholes that are relevant for place at the top level, that is, with bundles of properties and how these are different together than on their own. This is reflected in the definition of a point process as a stochastic model of irregular point patterns as a whole and without emphasis on individual points or realizations (Illian et al., 2008: 23). We are dealing here with a generative process that produces certain types of results, which means that we are still at the level of a stochastic process. This can be contrasted with the idea of a regression model, as often used in psychometric analysis. These are tools for predicting or explaining a variable from the information about other variables, that is, for relating them with each other (Mohr et al., 2021: 302). These types of models are therefore rather instrumental and often used pragmatically, reflecting the idea of artificially re-coupling what has once already been together, as we discussed further above. The reason why I chose point processes as a helpful analogy to illustrate some core ideas about processes of type (ii) is therefore precisely because point processes are formal representations of the generative process of a whole.

Conclusions

This article discusses a conceptual framework for the nexus of humanistic-geographical place and statistics. I first undertook a thorough assessment of the relevant place concept. This involved a discussion of two key components of place in the humanistic school: wholeness, which we considered in terms of both place itself and its investigation; and essentialism, which we conceived of in a statistical manner. On the basis of these considerations, we have outlined a dynamic structuralist conception of place, which is understood as a bond between self and world and which not only merely encompasses subjective and objective realms but amalgamates them into a whole. After a high-level discussion of where randomness can play a role in place, we turned to two fundamental issues: an adequate statistical mindset to be adopted towards place, and possible types of stochastic processes as potential statistical objects of investigation. I characterized the former with the idea of upward synergistic abstraction in contrast to the usual downward analytical decomposition. With regard to possible stochastic processes, we have focused on two place aspects: the concrete living of place and the resulting dormant structure that can be re-enacted at any time and remains in the mind. The current article is intended to provide a first and foundational basis for more concrete formulations of what I have termed platial synergistics.

In contrast to many existing quantitative and formal approaches to place, which tie the subject of place closely to the existing GIScience conceptual universe, I have deliberately avoided doing so. Examples of such approaches are Purves et al. (2019), who explicitly refer to the established core concepts of spatial information, Papadakis et al. (2016), who define place as space associated with functions, and Gao et al. (2013), who offer one of the few explicitly statistical place-related approaches from GIScience, but ultimately also refer to geometric primitives. Such approaches offer the great advantage of compatibility with the broader GIScience body of knowledge. The disadvantage that arises from this, however, is that a separate conceptual evolution regarding place would take place in parallel to that in human geography, which would lead to a further alienation between these two geographical branches. Since ultimately both GIScience and human geography mean the same object of study, the present article deliberately tries to refrain from such alienation and, in the spirit of Brachman (2020), to build the bridge by means of explicit reference to an established place concept from human geography and core ideas from GIScience (here: from spatial analysis).

Like the present article, some existing formal approaches to place avoid recourse to geospatial principles such as geometric coordinates. Scheider and Janowicz (2014) discuss a place reference system based on affordances and state that “place reference systems […] need to proceed from phenomena grounded in certain human abilities, and cannot take a detour based on space phenomena” (p. 100). The discussion in the present article echoes this statement and complements the work of Scheider and Janowicz by extending some of their core ideas to experience as understood in humanistic geography and to a statistical perspective. Another example of a treatment of place in GIScience that does not fit strictly into the established geospatial discourse universe is found in Winter and Freksa (2012), who define places contrastively. The authors posit “places as the units of (human) cognitive space” and point out that geometric boundaries do not play a role because “separation from the next unit is already done by identity” (p. 36). The attentive reader will have noticed that I have not once referred to boundaries in this article, because the metaphorical bundles of wire rope gain their identity from a particular evolving type of wiring (type (ii) processes) or the way they unbundle and interact (type (i) processes). My proposal for an overarching framework for statistical purposes thus ties in with existing formalizable place concepts in GIScience, but offers an approach that remains as strictly as possible within the conceptual bounds of an existing human-geographical place concept.

Since the present article provides a reference framework, the discussions I have offered would of course need to be expanded both in terms of further fundamental aspects and in terms of articulating operationalizations. One important topic that I left out of my discussion (primarily due to space constraints) is the link between place and probability theory. In the “Stochastic processes” section, I stated that statistical methods link empirical facts to hypotheses formulated in the form of probability distributions. However, the notion of probability is not uniform, and different notions of probability result in different reasoning frameworks. Arguably the prevailing and most familiar notion of probability in both human geography and spatial analysis is the frequentist one. However, it remains an open question as to whether this notion or one of the many others can provide a suitable framework for inferences about indeterminacy in a place context. For example, assuming that we can never have full knowledge of a domain like that of a case study’s place-making people, we may not be able to make qualified statements in the frequentist sense. In such a scenario, it might be more fruitful to develop a statistical apparatus for place based on the notion of qualitative probability (Delgrande et al., 2019) or some other probability notion. In the vein of Brunsdon (2017) arguing for a stronger position of Bayesianism in human geography, future research should discuss which form of probability is most suitable for the given context. Another current research strand that my article ties in with is the work on humanistic GIS presented in Zhao (2022). I deliberately did not deal with processes that concern the mediation of places into what could be called “place traces,” that is, reflections of place into data. Bo Zhao’s work focuses on GIS-mediated human experiences and, in conjunction with more general views on human–technology relations as offered by Verbeek (2008), can serve as a fruitful starting point for statistical perspectives on mediated places, especially in conjunction with the recent work of Mocnik (2023) on place representation.

The present article, with its broad formulation, offers numerous starting points for further research and for the development of statistical perspectives on place. The latter, by providing a way to rigorously reveal structures, can be a fruitful complement to the predominantly qualitative approaches to this subject. This work also contributes to the consolidation of the so far largely fragmented approaches to place within GIScience (see Merschdorf and Blaschke, 2018). Most importantly, this article can be hoped to serve as a fruitful basis for research into place in line with both GIScience and human geography, and thus to literally contribute to bridging between place and statistics—and possibly beyond with regard to additional formal and quantitative perspectives.