Space syntax and design

I take for granted that the design of the built environment involves principles that connect internalized repertoires of forms to functional effects; also, that the design imagination is strengthened as these principles – essentially theories that support descriptions and models of function – become explicit, are opened to criticism and test, and are enriched by research.

The first major engagement of space syntax with design is associated with the King’s Cross masterplan by Norman Forster in 1988 (Buchanan, 2006). The earliest alternative schemes by Foster evinced conviction regarding the placement, extent, and shape of the oval park in the middle – the proposals’ hallmark – but indicated uncertainty, or flexibility, regarding the treatment of the site between park and perimeter. In response, space syntax contributed to the design of the street network helping to flesh out four interlaced design aims: the elimination of the ‘urban hole’ associated with the King’s Cross goods yard, by making new connections between the neighbourhoods around it; the creation of an expanded ‘good address’ for new office buildings given the very narrow frontage on Euston Road; the balance between the connectivity of commercial and housing areas; and the creation of a coherent and lively urban environment to the benefit of all – not just the future inhabitants of the development.

Hillier (1993) compared the proposal to two hypothetical alternatives. First, the creation of an internal urban grid conceived independently of the surroundings and meeting the perimeter at arbitrary points. Second, the maximum inward linear extension of all major streets that are incident on the site perimeter. The technical device that allowed the spectrum of possibilities to be evaluated was the analysis of integration, or closeness centrality, with direction changes as the measure of distance, based on the axial map that represents the street network by the fewest and longest intersecting lines of sight and potential movement that cover all of its parts. The visualization and distribution of the integration core, the set of continuously linked most integrated lines of a system, acted as a key descriptor of spatial structure as it emerged from the interaction between design moves within the site and the pre-existing street layout around it.

The validity of integration analysis rests on multiple research findings. First, the distribution of integration is correlated with the distribution of pedestrian movement densities (Hillier et al., 1987, 1993; Peponis et al., 1989, 1997; Read, 1999) – the correlation holds after controlling for land uses (Özbil et al., 2011) or for development morphologies and densities (Berghauser Pont et al., 2019). By implication, street layouts support and distribute probabilities of co-presence and encounter. In doing so, they generate patterns of distributed attraction whereby commercial uses are not only aggregated around a hierarchy of central places but also extend linearly along central streets (Hillier, 1999; Scoppa and Peponis, 2015).

Second, urban street networks are almost universally characterized by a clear distinction between two sets of street lines: A small set of long lines, often intersecting at wide angles to create a primary global grid that traverses and connects the city in multiple directions; and a large set of shorter infill lines, often intersecting at right angles (Hillier, 2002). The interaction of these two scales of organization is reflected in the distribution of local integration – computed within a radius around each street element – and global integration – computed with regard to the system as a whole.

The relevance of integration analysis rests on two additional insights. The finding, above, that the distribution of pedestrian densities is correlated to integration measured according to path rotation (Turner, 2007) or direction changes, rather than metric length, suggests that cognitive factors are at play (Hillier and Iida, 2005). Length is associated with physical effort, while direction changes are associated with cognitive effort. Second, there is evidence (Hillier et al., 1987, 1993) that the distribution of movement is more strongly associated with integration when global integration is better correlated with local integration, or with the number of intersections along each street line. Local connectivity can be directly perceived or explored; global connectivity can only be grasped when allocentric and coordinated knowledge of environment is acquired. We can infer that layout intelligibility underpins the association between movement densities and syntactic structure. Indeed, the interface between the primary grid and the infill grids is the key factor governing intelligibility (Hillier, 2012).

The design of street layout is thus informed by a better understanding of what constitutes the spatial structure of the street network, what are its functional effects, and what is the underlying mechanism that underpins such effects. Space syntax helps to enrich the conceptualization of the configurational properties that design works with, as well as the effects design works towards.

King’s Cross illustrated how analytical theories can help to retrieve underlying structural principles from the study of precedents without copying the forms of the past. Lionel March (1998) contrasted space syntax and shape grammars according to their emphasis on necessity and freedom, respectively. King’s Cross suggests that syntax research accounts for the constraints within which architects can exercise imaginative freedom. This was the aspiration since the earliest papers by Hillier and Leaman (1974). In Foster’s (1997) words, space syntax bridges between intuition and analysis. Indeed, constraints are not the same thing as necessity. The purposeful exercise of freedom is more intelligible once constraints are understood.

The opportunity to transform the larger-scale pattern of centralities of the street network by designing one site is not always present. The contributions of space syntax to the redesign of Trafalgar Square, also by Foster, in the late 1990s and early 2000s (Buchanan, 2006), illustrate how a careful reading of circumjacent spatial structure can inform the elaboration of the internal spatial arrangement of a site, to the significant benefit of the life of place. But the question I want to address next is: even as space syntax has also been applied to the design of buildings, why do we not yet have an iconic project, equivalent to King’s Cross?

In buildings too, the intelligibility of space depends on the presence, the distribution, and the links between a small number of spaces – atria, corridors, courtyards, halls, and lobbies – configured so that no other space is more than a few direction changes away from them (Peponis, 2012). Furthermore, absent prior familiarity, exploration, and wayfinding are associated with syntactic connectivity and integration in a variety of building types, including hospitals (Haq, 2003; Haq and Zimring, 2003; Peponis et al., 1990) and museums (Choi, 1999; Tzortzi, 2015).

Identifying such underlying regularities, however, does not provide a sufficient basis for design. In building design, the generic human affordances of space associated with perception, movement, and encounter are harnessed to create affordances specific to social roles and organizational or institutional settings. In papers published in Environment and Planning B, spatial affordances have been studied in relation to interactions at work (Penn et al., 1999; Rashid et al., 2009; Wineman et al., 2014), informal learning in museums (Lu and Peponis, 2014; Peponis et al., 2004), visitor paths in museums (Choi, 1999), work culture in nursing units (Cai and Zimring, 2019), and informal supervision in restrictive settings (Peatross, 2001).

The outcomes found to be significantly correlated to spatial affordances also vary. They range from accelerated collaboration and organizational learning as evidenced by the reduction of the proportion of billable hours associated with problem conceptualization and design ideation for similar projects (Peponis et al., 2007), to the degree of publicly documented collaboration in research (Kabo et al., 2015), mortality rates in an ICU after controlling for medical condition, nurses’ qualifications, and other intervening variables (Ossmann, 2022), or pictorial theme recognition in museums (Lu and Peponis, 2014).

Between outcomes of interest, observable phenomena, and affordances, the gap can be large. In buildings, however, bridging the gap is facilitated by more sensitive and rich descriptions of the morphologies of inhabitation. We have understood spatial cognition and wayfinding in hospitals, and patterns of exploration in museums, by studying samples of actual paths, rather than taking snapshots of the distribution of moving people. We have learned about organizational cultures by recording the spatial location of face to face-conversations against the background of detailed maps of space occupancy and also by mapping the networks of interaction reported in questionnaires. And we acquire a better grasp of informal learning by using eye tracking in relation to the study of paths and stops in museums. As we refine our descriptions of the morphologies of inhabitation, so we can also refine our theories of spatial functions and, more important, our definitions of design intent.

Only some of the cornucopia of interdisciplinary studies of the human, social, and cultural functions of buildings are likely to be directly relevant to a particular design program. One more reason to ask whether research findings converge towards general propositions. I think that the answer is positive but not simple. The propositions I have in mind do not bear on a correlation between syntactic properties and directly observable phenomena such as the distribution of movement densities in urban space. We must also take into account the ways of seeing and understanding that mediate the social function of space. I have recently argued (Peponis, 2024a) that we may distinguish two different codes by which people make sense of the spatial syntax of environment as they inhabit it. Distributive codes are about knowing where people and things are, and how to access them. Correlative codes are about discovering what is available in an environment, and developing an enriched and open-ended understanding of the relationships that things and people can enter into. Based on research so far, correlative codes seem to require that the global spatial structure of a building be intelligible, thus allowing attention to be directed to local path choices motivated by the desire to seek or avoid opportunities for interaction or other forms of engagement. Also, that the most integrated spaces be richly invested with activities other than circulation, and visually connected with circumjacent primary use spaces, so that movement can work to generate co-awareness, contact, and communication, thus balancing the opportunities for seclusion offered in less integrated areas.

The earliest applications of space syntax to the study of buildings were associated with the search for the invariant associations between the socially constructed labels through which we identify and characterize individual spaces – for example, ‘living room’, ‘bedroom’ and, ‘kitchen’ – and the rank order of depth of such spaces from the entrance, or their centrality within building plans of a particular building type (Hillier et al., 1987; Hillier and Hanson, 1984; Markus, 1987). In short, the emphasis was upon invariant socio-spatial relations. Hanson (1998) pushed the study of constraints towards an exploration of cultural transformations and the limits and meaning of design choices.

As the field of space syntax continues to mature, future applications to building design are likely to be associated with invention and innovation regarding the specification of design ends, rooted in a deeper understanding of the morphologies of inhabitation. They are also likely to be associated with the enrichment of design means rooted in the ever more refined description of the human and social affordances of the built space. Arguably, making the dialogue between means and ends more transparent is one of the contributions of space syntax (Peponis 2024b). In pursuing this dialogue, space syntax may lean towards what Lionel March called ‘freedom’ but will do so while continuing to affirm that function is intrinsic to form, and that research should address the work that the syntax of space performs within culture.