Queering Complexity Using Multi-Agent Simulations

Abstract

In this paper, we demonstrate how multi-agent simulations of gender and sexuality-based marginalization can help us understand gender and sexual experiences as complex, emergent, multi-level phenomena, which involve dynamic interactions between individuals, groups and institutions. We present Flocking QT Stories, a multiagent-based simulation that illustrates how structural (macro-level) phenomena such as gender and sexuality-based marginalization and resilience can manifest through individual-level interactions between computational agents. We then present illustrative cases from an ongoing co-design study with people with lived or professional scholarly experience of gender and sexuality-based marginalization. Our analysis reveals how participants interacted with Flocking QT Stories through turning (Ahmed, 2006) their attention toward marginalized agents, and engaging in multi-level reasoning (Hostetler, Sengupta & Hollett, 2018; Wilensky & Resnick, 1999) to make sense of the audio stories embedded in the simulation as well as the macro-level, emergent behaviors resulting from interactions between individuals (boids), and between individuals and institutions.

Full text

Queering Complexity Using Multi-Agent Simulations
Dylan Paré, Marie-Claire Shanahan & Pratim Sengupta
dylan.pare@ucalgary.ca, mcshanah@ucalgary.ca, pratim.sengupta@ucalgary.ca
University of Calgary
Abstract: In this paper, we demonstrate how multi-agent simulations of gender and sexuality-based
marginalization can help us understand gender and sexual experiences as complex, emergent, multi-
level phenomena, which involve dynamic interactions between individuals, groups and institutions.
We present Flocking QT Stories, a multiagent-based simulation that illustrates how structural
(macro-level) phenomena such as gender and sexuality-based marginalization and resilience can
manifest through individual-level interactions between computational agents. We then present
illustrative cases from an ongoing co-design study with people with lived or professional scholarly
experience of gender and sexuality-based marginalization. Our analysis reveals how participants
interacted with Flocking QT Stories through turning (Ahmed, 2006) their attention toward
marginalized agents, and engaging in multi-level reasoning (Hostetler, Sengupta & Hollett, 2018;
Wilensky & Resnick, 1999) to make sense of the audio stories embedded in the simulation as well
as the macro-level, emergent behaviors resulting from interactions between individuals (boids), and
between individuals and institutions.
Keywords: multi-agent modelling, complex systems, gender, sexuality
Introduction
In this paper, we demonstrate how interacting and reasoning with multi-agent simulations of gender and sexuality-
based marginalization can help us understand gender and sexuality experiences as complex, emergent, multi-level
phenomena, that involve dynamic interactions between individuals, groups and institutions. In complex systems,
interactions between smaller or individual components of the system give rise to collective, system-level behaviours
that are often difficult, counter-intuitive or even impossible to predict simply by thinking about the behavior of each
individual component (Holland, 1998; Mitchell, 2009). We offer a complementary approach for designing learning
environments to support critical conversations on gender and sexuality that brings together scholarship on queer and
trans theories (Ahmed, 2006; Butler, 1990; Lane, 2009) and complex systems education using multi-agent
computational simulations (Wilensky & Resnick, 1999; Hostetler, Sengupta & Hollett, 2018). We present Flocking
QT Stories, a multiagent-based simulation that illustrates how structural (macro-level) phenomena such as gender and
sexuality-based marginalization and resilience can manifest through individual-level interactions between
computational agents. We then present illustrative cases from an ongoing co-design study with people who have lived
experience and/or professional scholarly experience of gender and sexuality-based marginalization. Our analysis
reveals how participants interacted with Flocking QT Stories through turning (Ahmed, 2006) their attention toward
marginalized agents, and engaging in multi-level reasoning (Hostetler, Sengupta & Hollett, 2018; Wilensky &
Resnick, 1999) to make sense of the individual, audio stories embedded in the simulation as well as the macro-level,
emergent behaviors resulting from between individuals (boids), and between individuals and institutions.
Theoretical Background: Synergies between Complexity and Queer Theories
Our theoretical framework is premised on synergies between several foundational constructs proposed by queer
theorists such as the heterosexual matrix (Butler, 2006/1990), queer orientations (Ahmed, 2006), as well as Lane’s
(2009) call for trans and queer studies to engage with a fundamentally more complex, new materialist biology. We
propose that both Butler’s notion of the heterosexual matrix and Lane’s notion of body-becoming can be understood
in light of mechanisms of emergence, i.e., how unintentional system-level behaviors emerge from individual-level
interactions (Holland, 1998).
For example, Butler’s (2006/1990) notion of the heterosexual matrix helps us understand how the hegemonic
discursive epistemic practices enacted in our everyday interactions reify heterosexuality as normative. Butler further
contends that the everyday reproduction of the “compulsory practice of heterosexuality” (Butler, 2006/1990, p. 208),
typically remains hidden from the view of individuals. At the system level, the emergent behaviour identified in this
model is represented in the pattern or social practices of disciplining bodies, or the way that bodies are made to “cohere
and make sense” within a linear, causal framework of assumed sex-gender-sexuality. Butler (2006/1990)
conceptualizes power as discursively produced in interactions at social and institutional levels - not as a centralized

power of control - a point echoed by Risman (2004, 2018), who positioned gender as a social structure, in which both
material and cultural (ideological) processes are at work within and across individual, interactional, and macro levels.
Lane (2009) further advances the argument for re-positioning gender and sexuality as emergent rather than
deterministic. Lane (2009) critiques Butler for emphasizing “the fixity of biological materialities of sex compared
with the flexibility of gender” (p. 141). Lane argues that instead of viewing the body as constraining, fixed, and given,
we must view the body as a dynamic, transformative process of ‘body becoming’ (Lane, 2009, p. 141), where social,
material and biological aspects of experience interact with each other fluidly.
Along similar lines, Ahmed (2006) argues that queerness is not simply an identity defined by one’s desire,
but is a lived experience of “the everyday work of dealing with the perceptions of others, with the ‘straightening
devices’ and the violence that might follow when such perceptions congeal into social forms” (p. 107). Further, Ahmed
(2006) demonstrates how straightness is not the default or “normal” identity but is also produced. In this way,
becoming straight, “as a direction (taken) toward objects and others, is compulsory. In other words, subjects are
required to ‘tend toward’ some objects and not others as a condition of familial as well as social love” (p. 85).
Connecting each of these orienting gestures is the way that relationships with people, spaces, and objects are all
implicated in the (re)production of cisheteronormativity.
As our discussion so far shows, critical gender and sexuality scholars have positioned the body, the
individual, as well as interactions with others and institutions as complex and emergent rather than linear and
deterministic. As a corollary, we believe that challenges in public understanding of these issues can be explained in
light of conceptual difficulties in understanding complex emergent systems that arise viewing non-linear, emergent
processes as deterministic (e.g., Sengupta & Wilensky, 2009; Wilensky & Resnick, 1999). In the context of gender
and sexuality, this means that the emergent, fluid nature of gender and sexuality is viewed instead as biologically
predetermined. That is, instead of viewing the social, historical and institutional forces and interactions – both implicit
and explicit – that shape our gender and sexual experiences and identities – an emergent view – we tend to
conceptualize gender and sexuality as an individual “trait”, that is predetermined biologically – a deterministic view.
In the literature on complex systems education, such conceptual challenges have been noted as examples of
slippage between levels (Sengupta & Wilensky, 2009; Wilensky & Resnick, 1999), which happens when people –
experts and newcomers to disciplines alike – incorrectly assign individual-level attributes (e.g., individual cars moving
forward) to the aggregate-level phenomenon (e.g., traffic jams moving forward) thereby leading to an incorrect
prediction (e.g., in reality, traffic jams propagate in the backward direction). However, a substantial body of research
now shows that multi-agent computational models can provide learners with an intuitive pathway for reasoning about
emergent outcomes by enabling learners to think like an agent and, consequently, bootstrap their embodied and
intuitive knowledge of individual-level actions and interactions (Dickes, Sengupta, Farris, & Basu, 2016; Sengupta &
Wilensky, 2009). And finally, although the intersection of critical theory and complexity has received relatively little
attention, recent scholarship suggests that multi-agent models can be productively used to unsilence critical
conversations about race (Hostetler, Sengupta & Hollett, 2018). Our current work seeks to further expand this field
by identifying critical issues in gender and sexuality as a new context for complex systems education and by
illustrating how multi-agent computational models can serve a generative and productive computational medium for
representing these issues in public settings.
The Simulation: Flocking QT Stories
In our simulation, there are four types of computational agents: normative and non-normative boids or bird-droids
(Reynolds, 1987), and two types of institutions: normative and non-normative (see Figure 1). Proximity to normative
boids and institutions “drains” energy from the non-normative boids, and proximity to other non-normative boids and
institutions increases their energy. Each non-normative boid “carries” an audio story - a first person account of gender
and/or sexuality-based marginalization and resilience. These stories were recorded by local people who volunteered
to contribute a short 2-minute audio narrative of their experiences of gender and sexuality-based marginalization or
resilience. All of the boids move according to Reynolds’ (1987) flocking algorithm. In Reynolds’ algorithm, the rules
obeyed by each boid are alignment, separation, and cohesion. Alignment means that a boid tends to turn so that it is
moving in the same direction that nearby boids are moving. Separation means that a boid will turn to avoid another
boid which gets too close. Cohesion means that a boid will move towards other nearby boids (unless another boid is
too close). In addition to these forces that are always in play, when a non-normative boid’s story is activated, all the
non-normative boids’ movements are also affected in part by the frequencies of the audio file. Higher-frequency
sounds affect the cohesive motion, lower-frequency sounds affect the separation, and medium-frequency sounds align
their directions with neighboring boids. The overall effect is that the movement pattern of the non-normative boids is
visually distinct from the normative boids while a story is being played, causing the former to vibrate while they are
also flocking. This also helps create a visual contrast between a flock of non-normative boids that are moving together
with vibrations, and a non-normative boid that is stagnant because of being isolated from other non-normative boids

and due to being positioned very close to a normative institution that drains its energy. This contrast in movement, we
posit, can help orient peoples’ attention to representations of marginalization.
People can interact with the simulation by clicking on (or touching) a non-normative boid in order to listen
to the story carried by it, which in turn affects the movement of the other non-normative boids based on the rules
explained above. They can also use the slider shown in Figure 1 to change the ratio of the normative to the non-
normative boids. People’s interactions with the simulation are mediated by a facilitator (one of the authors or other
members of the research team affiliated with the project). Engaging in conversations with the facilitator as well as
with other participants is an important element of people’s experiences with the simulation. The audio stories recorded
in the simulation include a wide range of experiences of gender and sexuality-based marginalization and resilience,
with a focus on everyday interpersonal and institutional interactions. These include stories such as: bisexual erasure
in queer and straight spaces, growing up queer in a small town, experiences of homophobia in a Catholic school and
other religious spaces, facing limiting gendered expectations for women in male-dominated workplaces, and the
influences of hegemonic masculinity in the academy, etc. The narrators are people across a wide range of gender and
sexual identities, including people who identify within LGBTQ identities as well as cisgender and heterosexual people
— all of whom have experiences of gender and/or sexuality-based marginalization.
Figure 1. Screenshot of the Flocking QT Stories multi-agent simulation
Some interesting emergent patterns in the simulation include non-normative boids escaping normative
institutions as they find a sufficient number of flock mates, and conversely, getting trapped (i.e., becoming immobile
due to losing energy) when they are too close to normative boids and/or normative institutions while being separated
from non-normative flocks. These emergent behaviors represent resilience (in the first case) and marginalization (in
the second case). Furthermore, as our analysis will reveal, these simulated, emergent representations open up the
interpretive space for learners and members of the public who interact with the simulation by inviting them to draw
parallels from their own experiences and the experiences shared in the pre-recorded stories carried by the boids, in
order to interpret what such behaviors might represent. This is a particular affordance of agent-based models and
representations, specifically in contexts that involve critical theoretical issues (e.g., racial and ethnic segregation)
(Hostetler, Sengupta & Hollett, 2018). Computational agents can act as transitional objects (Papert, 1980), i.e., people
can project their own experiences onto the agents as they reason about the behaviors of the agents. Building on this
idea, Hostetler, Sengupta & Hollett (2018) noted that when pre-service teachers use multi-agent simulations of
ethnocentrism, they can engage personally with controversial issues while maintaining sufficient epistemic and
affective distance from the discourse.
However, rather than ascribing the epistemic power of this work simply to the protean nature of
computational agents, Sengupta, Dickes & Farris (in press) noted that central to supporting such critical discourse
using multi-agent simulations is the design of activities that help students productively recontextualize generalized
computational algorithms and visual representations. For example, in Hostetler, Sengupta & Hollett’s (2018) study,
using the Racial Dot Map (which shows residential segregation on a map of the USA) alongside the simulation of
ethnocentrism (which relies on a generalized algorithm grounded in Game Theory) was essential for pre-service
teachers being able to engage in critical discourse about race and inequality. Similarly, our goal in integrating personal
narratives of gender and sexuality with the simulation is to engage the participants (including learners and members

of the public) deeply with first-person narratives of gender and sexuality-based marginalization and resilience while
interacting with the Flocking QT Boids simulation. These stories serve as scaffolds for helping participants understand
the simulation’s emergent patterns of harm and support in the context of gender and sexuality-based marginalization.
They also help the participants differentiate the behavior of normative and non-normative boids, given that only the
non-normative boids carry stories.
Research Question
In this study, we ask the following research question: How do people with lived experience and/or professional
scholarly experience of gender and sexuality-based marginalization – ranging from critical gender and sexuality
scholars to queer undergraduate students - explain emergent forms of marginalization and resilience through
interacting with the simulation?
In asking this question, our research objectives are two-fold: a) to deepen connections between scholarship
on complexity education and multi-agent models, and critical scholarship on gender and sexuality; and 2) develop
this synergy by centering experiences of gender and sexuality based marginalization across all genders and
sexualities in a manner similar to McWilliams’ (2016) work to queer participatory action research by fundamentally
challenging categories around gender and sexuality.
Methods
As part of an ongoing co-design study with people with lived experience and/or professional scholarly experience of
gender and sexuality-based marginalization, we interviewed two groups (dyads) of participants as they interacted with
the simulation. The study setting is DigiPlay (Sengupta & Shanahan, 2017), a public walkway located in a Canadian
university where visitors can interact with open source, multi-agent simulations of complex systems using 80 inch
touchscreens. One dyad of participants, Chuli and Safa, are university level professors and instructors engaging in
critical scholarship of gender and sexuality. The second dyad are two undergraduate, university students, Sophia and
Chuck, who identified themselves as queer (1). We chose these participants because we were interested in
understanding how people with lived experience and people with scholarly experience of gender and sexuality-based
marginalizations would make sense of the multi-agent simulation. Our decision to interview dyads of participants who
are familiar with each other is based on recent scholarship which highlights the value of conversations with friends
for engaging in critical and conceptually challenging discussions, including issues of gender and sexuality-based
marginalization (Paré, Sengupta, Windsor, Craig & Thompson, 2019; Takeuchi, 2016).
The videos were viewed individually by each researcher and codes were developed and discussed using the
constant comparative method (Corbin & Strauss, 1990; Glaser, 1965). During the first part of this process, known as
open-coding, codes were discussed in relation to data. Our theoretical focus was on identifying episodes – i.e., turns
of conversations – that involved multi-level reasoning. That is, we were specifically interested in identifying
explanations that involved both individual level descriptions - such as personal narratives, and agent-level rules – and
interactions between individuals, among individuals and institutions or larger social forces. During the second part of
this process, known as axial coding, each code was discussed in relation to others. This process helped us refine themes
by identifying the different examples of resilience and marginalization that the participants brought up in their
explanations. Overall, two themes emerged in our findings: the use of “turning” (Ahmed, 2006) and multi-level
reasoning to identify and explain examples of resilience and marginalization (Theme 1), and multi-level perspectives
for extending the simulation in order to deepen our understanding of gender-based resilience and marginalization
(Theme 2).
Findings
Theme 1: “Turning” and multi-level reasoning to explain resilience and marginalization
In our first illustrative case, we follow an ongoing discussion between Sophia and Chuck in which they show concerns
about the on-screen interactions of a non-normative boid that had become stagnant (Figure 2). The stagnant boid was
isolated and seemingly trapped due to its immediate proximity to a normative institution, and distance from other non-
normative boids or non-normative institutions. Orienting their attention to the stagnant boids was a common
experience for all participants in our data to date (including Chuli and Safa, our second pair of participants). This is
an act of turning (Ahmed, 2006) or tending their attention toward a computational representation of marginalization.
Sophia frames her observation of these on-screen behaviours by explaining the boid’s experience as comparable to
being the only queer person in a small town (Excerpt 1).

Turn 1 Sophia (She/Her & They/Them): [Pointing to a transparent, non-normative boid that is not moving with
harm lines linking it to a normative institution square and an opaque non-normative boid linking with the
transparent boid with support lines] Look at this one trying to get well with the institution draining it and the
one best friend.
Turn 2 Chuck (They/Them): Oh, yeah, same with the, like, this one. [Points to another non-normative boid that
is stagnant and linked to a normative institution with harm lines] It's been stuck there for like basically the
whole entire time. And to help it get better the other purple triangles have to interact with that institution. It's
just really hard.
Turn 3 Sophia: Well, and the. [points back at a stagnant, isolated non-normative boid next to normative
institution] It's also. It's like a small town. It's like it's isolated from every. Like, over here [motions to area with
a cluster of normative and non-normative institutions and to an area with a flock of non-normative boids] there's
more chance of you running into more people and institutions, but there's like [motions back to the isolated
non-normative boid]
Turn 4 Sophia: Interesting too, you'd think that when they're that strong [points to large flock of non-normative
boids], they could avoid the normative institutions [points to a normative institution], but they're not avoiding
them probably because they have to be. They have to participate in them.
Excerpt 1. Discussion of effect of small towns on queer people
Figure 2. Screenshot of the Flocking QT Stories simulation showing an example of a stagnant non-normative boid
(triangle with pink outline) trapped near a “normative” institution (white square)
This interaction demonstrates how Sophia and Chuck make sense of non-normative boids who become
“stuck” in institutions. This is a form of multi-level reasoning, because although the participants were focused on
interactions between two agents, they also make a connection to a macro-level observation of queer experience - the
effect of small towns on queer people, and how difficult it can be to find support to survive. This form of multi-level
discourse continues later in their interaction, when Sophia then notices that when non-normative boids are moving
together, they can better survive the effects of normative institutions, although they cannot avoid the harm altogether.
She connects this phenomenon back to personal experience, asking Chuck about the kind of scenario they would prefer
(Excerpt 2).
Turn 1 Sophia: It's the same here [points to two non-normative boids that are next to a normative institution
with harm lines between each boid and the institution and a support line connecting the non-normative boids].
One friend is not enough to combat an institution.
Turn 2 Sophia: [Directing question to Chuck] Would you rather be, like, in communities like this where you
have more interaction with people but there's lots of harm in it [pointing to large flock of normative and non-
normative boids] or in like a small, threesome with no harmful interaction that's like marginalized from society?
[gestures with fingers making closing motion to indicate a smaller group]
Turn 3 Chuck: [Points to a small flock of all non-normative boids] Yeah, I think that one. [pointing to the small
group of three non-normative boids]

Turn 4 Sophia: Well, yeah. [both laugh] Like, who wants that one? [check video to see which is indicated]
The like five or six geometry shape of total safety. [Gestures to bring hands together, interlocking ends of
fingers].
Excerpt 2. Discussion of queer experiences of collective resilience and isolation
In Excerpt 2, during Turns 1 and 2, Sophia’s explanation again provides evidence of multi-level reasoning.
For example, in Turn 1, she points to interactions between a small group of agents - two non-normative boids and a
normative institution. In Levy and Wilensky’s (2008) terms, this is evidence of a mid-level observation - as it is in
between the individual level (a single boid) and the macro-level (e.g., a large flock). Yet, the power of this mid-level
noticing, as Levy and Wilensky (2008) argued, is that these interactions can reveal to the observer important
mechanisms that can be used to explain and understand larger scales of complexity. In this case, Sophia, in Turn 1,
makes an important inference based on this observation: “One friend is not enough to combat an institution”. This is
a poignant realization, as Sophia then continues this line of reasoning in Turn 2 as she notices a larger flock and
compares the interactions between boids in that flock with her previous observations. She positions her observations
in the form of a question to Chuck: Which community would you rather be in? The premise of her question is her
observation of the harm and resistance to harm that results from being in a mixed, larger flock, and a hypothetical
comparison with a smaller, mid-level flock with only non-normative boids. Chuck responds that they would rather be
a part of this smaller, non-normative flock, and as they are responding, such a flock appears in their visual field, to
which Chuck points (Turn 3). Sophia then recognizes this new, non-normative flock on the screen as “five or six
geometry shape of total safety” (Turn 4). Sophia and Chuck are now in consensus: they would rather see themselves
as part of a small non-normative flock, which even though it would be “marginalized from society” (Turn 2), would
still keep them safe, i.e., away from harmful interactions that may take place when the flock is larger and contains
both normative and non-normative boids.
What is evident from the analysis so far is that Sophia and Chuck have framed the experience of harm in
terms of interactions between different types of agents in the simulation: normative boids, non-normative boids, and
institutions. This is not surprising given that this was one of the design goals of the simulation from our (the designers')
perspective. Sophia then further expands on her understanding of harm by proposing that a new kind of agent - allies
- be introduced in the simulation that do not drain energy from non-normative boids. She mentions her idea of this
new type of boids because she is concerned for the extremely marginalized and isolated boid that is still trapped near
a normative institution (see Excerpt 1, where she first notices the boid). She connects this boid’s experience to a story
in the simulation, which she also recognized to be her sister’s voice. At the same time, Chuck also considers how
changing the number of institutions could shape individual and group experiences (Excerpt 3). In Turn 3 (Excerpt 3),
Sophia connects an individual story in the simulation (her sister’s story) to the boid interaction in the simulation of
the isolated boid, and wonders about the extent to which allies could help in this context. She uses her sister’s story to
hypothesize about how more non-normative boids (or other queer people as explained in the story) are needed for
support, and that allies might have a role to play where a non-normative boid (queer person) is isolated in a small
town, but that allies may not be able to provide adequate support.
Theme 2: Multi-level perspectives on extending the simulation
Turn 1 Sophia: I'd be very interested to see what would happen if you added, like, actual quality allies to this.
Like, um, like, 'cuz like, um, I can't remember the pseudonym she used, but my sister talked about being among
supportive family [makes hand gesture closing fingers to indicate a small group], but just not having any other
queer people. So, it's not like, but the institutions were like this [points to normative institution]. I don't know,
it's just like a different combination of like, trying to get her support for her queerness but not having other
queer people, but having allies.
Turn 2 Chuck: [Points to slider to change ratio of normative to non-normative boids to suggest a slider that
would change the ratio of institutions] You mean if we could like up the level of non-normative institutions as
opposed to normative ones? In here so like they're all purple?
Turn 3 Sophia: Oh yeah, that would be interesting. Or just like human allies too. Like, boid allies that aren't
automatically going to drain and do harm. [points to small flock where a normative boid has harm lines draining
energy from multiple non-normative boids around it] So, rare. [both laugh] Rare to find true allies, but, um, I
wonder if support from an ally would be equally nourishing or probably less. [looks to Chuck] Probably less if
her story was indicating. 'Cuz she didn't feel able to be herself despite having some pretty close, strong allies.

Excerpt 3. Introducing allies
In the Excerpt 3, Sophia suggests adding “allies” - a new type of agents - that would represent people who do not
themselves identify as queer but are supportive of queer people. Sophia drew from her personal experience - her sister
- who also identifies as queer, and talked about the value of being with a small group of allies such as a supportive
family members (Turn 1). Chuck wondered if this would entail adjusting the ratio of normative to non-normative
institutions to represent greater support for queer people by increasing the proportion of non-normative institutions
(Turn 2). Sophia followed up by mentioning that she was thinking of introducing human allies - i.e., a new type of
normative boids that do not drain energy from the non-normative boids (Turn 3).
Turn 1 Chuli (She/Her): So, you know, because there is a connection between institutions, people, and
ideology. Right, so, you know, how ideology or um, structural discourse gets internalized and so maybe if there
are like, I don't know, another shape for ideology or another sort of
Turn 2 Safa (She/Her): Something like a circle, in a different colour. But it takes over. I don't know.
Turn 3 Safa: I was thinking like a circle in a different colour or it takes over, I don't know. which is a...
Turn 4 Chuli: Yeah, the institution?
Turn 5 Safa: Yeah.
Turn 6 Facilitator: Okay, so circles that move, are you thinking?
Turn 7 Safa (She/Her): Yeah, 'Cuz I feel like, so, ideologies are like institutions. You have the shape of a
circle, like a coloured circle, not a full circle, that goes and you can see like to some extent, interactions with
the institution or with the person. It's like a little bit of a portion of it comes over or it takes over and changes
the colour of the institution, you know. Maybe something like that.
Turn 8 Chuli (She/Her): I think ideas could change spaces, but you need to have bodies, right, for those ideas
to be present. Like, you know, you can have one [facilitator's name] or one [Chuli] and that does only so much.
I mean [university name] as an institution is not changing because of our presence because we are not enough.
We have our voices, but who's hearing our voices is important. So I think what you have [indicating the
simulation] is very powerful in terms of the support network because being heard by other people and having
people hear you, you know, at your level. But I, yeah there's something about. So as we change the non-
normative, there were more normative boids. Is there a mechanism for normative boids to change colour?
Excerpt 4. Introducing Institutions
In a similar vein, Chuli and Safa (Excerpt 4) wanted to explore deeper forms of criticality in the form of
connections between “institutions, people, and ideology” (Chuli, Turn 1). As Chuli explained in Turn 1, her goal in
proposing a new form of agents - ideologies - was to make explicit how “structural discourse gets internalized”. In
Turn 7, Safa noted that the process of internalization could be visually represented by interactions between institutions
and ideologies, as institutions take on the shape or color of ideologies, and that in turn would have an impact on the
movement of non-normative boids around them. For Chuli, ideas are embodied through people and institutions, which
in turn should be seen as “networks” of people (Turn 8). She noted that simply having queer presence in an institution
does not automatically lead to the institution transforming itself: “We have our voices, but who's hearing our voices
is important”, she noted (Turn 8). This is also evidence of turning (Ahmed, 2006) attention toward experiences of
marginalization, but at the institutional level: Chuli argues that whether or not queer voices are heard within an
institution shapes the institutional orientation towards or away from queerness. She suggested that by making such
interactions visible, we might better understand the hidden mechanisms through which gender and sexuality are
experienced in institutional settings.
Summary and Discussion
Overall, we found that both dyads of participants used multi-level reasoning during their interactions with the
simulation in order to identify and explain examples of gender and sexuality-based marginalization and resilience. For
example, participants positioned harm as an emergent phenomenon that arose from interactions between different
kinds of agents in the simulation and reasoned about minimizing harm specifically for the most marginalized agents
in the simulation. Participants also proposed changes to the model to reason about adding new types of agents - allies
and ideologies - to reason about how individuals, institutions, and ideologies interact to produce emergent behaviours
such as institutional and social change, and queer marginalization. In each of the excerpts we reported, it was clear
that the participants were turning (Ahmed, 2006) their attention toward marginalized agents in the simulation.

However, this was not simply an individual-level noticing; the well-being, marginalization and resilience of queer
individuals were clearly positioned in light of interactions with other individuals (e.g., allies), institutions and
ideologies. In other words, instead of assuming that these are only individual level experiences - which would have
been an example of slippage between levels (Wilensky & Resnick, 1999) - participants were able to reason about the
interactional and emergent nature of such experiences using the simulation. This in turn illustrates that the experience
of taking on a queer orientation (Ahmed, 2006) involves thinking about emergence, an experience that is far more
complex than re-orienting our attention to an individual, that involves multi-level reasoning. To this end, we believe
that our analysis shows that Flocking QT Boids simulation can serve as an effective scaffold for engaging in such
reasoning and conversations.
(1) We use the term queer in this paper as a short-form to indicate that the participants identify within the LGBTQ+ spectrum of
non-normative genders and sexualities. This term, however, may not reflect the terminology that a participant uses to describe
themselves and does not necessarily describe their specific gender and sexual identities.
References
Ahmed, S. (2006). Queer phenomenology: Orientations, objects, others. Durham, N.C.: Duke University Press.
Butler, J. (2006). Gender trouble: Feminism and the subversion of identity. New York, US: Routledge. (Original work
published 1990)
Corbin, J. M., & Strauss, A. (1990). Grounded theory research: Procedures, canons, and evaluative criteria. Qualitative
sociology, 13(1), 3-21.
Dickes, A. C., Sengupta, P., Farris, A. V., & Basu, S. (2016). Development of mechanistic reasoning and multilevel
explanations of ecology in third grade using Agent‐Based models. Science Education, 100(4), 734-776.
Glaser, B. G. (1965). The constant comparative method of qualitative analysis. Social Problems, 12, 436-445.
Holland, J. H. (1998). Emergence: From chaos to order. Cambridge, MA: OUP Oxford.
Hostetler, A., Sengupta, P., & Hollett, T. (2018). Unsilencing critical conversations in social-studies teacher education
using agent-based modeling. Cognition and Instruction, 36(2), 139-170.
Lane, R. (2009). Trans as bodily becoming: Rethinking the biological as diversity, not dichotomy. Hypatia, 24(3),
136-157.
Levy, S. T., & Wilensky, U. (2008). Inventing a "mid level" to make ends meet: Reasoning between the levels of
complexity. Cognition and Instruction, 26(1), 1-47.
McWilliams, J. (2016). Queering participatory design research. Cognition and Instruction, 34(3), 259-274.
Mitchell, M. (2009). Complexity: A guided tour. Oxford University Press.
Papert, S. (1980). Mindstorms: Children, computers, and powerful ideas. New York, US: BasicBooks.
Paré, D., Sengupta, P., Windsor, S., Craig, J., & Thompson, M. (2019). Queering virtual reality: A prolegomenon. In
Sengupta, P., Shanahan, M-C., & Kim, B. (Eds). Critical, transdisciplinary and embodied approaches in
STEM education. (pp. 307 - 328). Springer, Cham.
Reynolds, C. (1987). Flocks, herds and schools: A distributed behavioral model. Proceedings of the 14th Annual
Conference on Computer Graphics and Interactive Techniques, 25-34.
Risman, B. J. (2004). Gender as a social structure: Theory wrestling with activism. Gender & society, 18(4), 429-450.
Risman, B. J. (2018). Where the millennials will take us: A new generation wrestles with the gender structure. Oxford
University Press.
Sengupta, P., Dickes, A.C., & Farris, A. (in press). Voicing Code in STEM: A Dialogical Imagination. MIT Press.
Cambridge: MA.
Sengupta, P., & Shanahan, M. C. (2017). Boundary play and pivots in public computation: New directions in STEM
education. International Journal of Engineering Education, 33(3), 1124-1134.
Sengupta, P., & Wilensky, U. (2009). Learning electricity with NIELS: Thinking with electrons and thinking in levels.
International Journal of Computers for Mathematical Learning, 14(1), 21-50.
Takeuchi, M. A. (2016). Friendships and group work in linguistically diverse mathematics classrooms: Opportunities
to learn for English language learners. Journal of the Learning Sciences, 25(3), 411-437.
Wilensky, U., & Resnick, M. (1999). Thinking in levels: A dynamic systems perspective to making sense of the world.
Journal of Science Education and Technology, 8(1), pp. 3-19.