Trends in

Cognitive Sciences
Opinion

Harnessing Visual Imagery and Oculomotor

Behaviour to Understand Prospection
Federica Conti 1,2,* and Muireann Irish 2,*

Much of the rich internal world constructed by humans is derived from, and Highlights
experienced through, visual mental imagery. Despite growing appreciation of Visual mental imagery is heavily impli-
visual exploration in guiding episodic memory processes, extant theories of cated in episodic retrieval; yet, its
contribution to future-oriented forms
prospection have yet to accommodate the precise role of visual mental imagery
of cognition remains poorly understood.
in the service of future-oriented thinking. We propose that the construction of
future events relies on the assimilation of perceptual details originally experienced, The visuo-oculomotor system is ideally
and subsequently reinstantiated, predominantly in the visual domain. Individual positioned to support the core memory
network in the simulation of future
differences in the capacity to summon discrete aspects of visual imagery can
experiences, suggesting shared under-
therefore account for the diversity of content generated by humans during future lying functional capacities.
simulation. Our integrative framework provides a novel testbed to query alterations
in future thinking in health and disease. Process-Specific Alliances between
the memory and visuo-oculomotor
systems likely support discrete aspects
of imagery reinstantiation during event
The Prospective Brain construction.
Much of human daily life consists of anticipating and preparing for what lies ahead. This so-called
prospective bias is evident in our everyday choices, whether imposed by immediate needs Altered visual mental imagery in clinical
populations provides a unique window
(e.g., thinking about our next meal) or more long-term decisions (e.g., choosing a career path
to understand the intersection between
or romantic partner). Imagining the outcomes of actions and simulating possible alternatives the memory and visuo-oculomotor sys-
enable us to rule out undesirable consequences and move progressively towards our goals in tems during future thinking.
a directed manner [1,2]. Although widely accepted as the content or output of future simulation
(see Glossary), the precise contribution of visual mental imagery to this process remains poorly
understood. That mental imagery should be central to human cognition can be traced back to
Aristotle’s prescient observation that ‘the soul never thinks without a phantasma’ [3]. Humans
largely apprehend, perceive, and experience the world through the visual system, the content
of which is recapitulated via mental imagery in the form of episodic memories. While the fields
of vision science and memory have largely proceeded in parallel, recent approaches suggest
that we might now be able to unite these disciplines [4].

Several converging lines of evidence suggest that visual imagery should, and does, contribute
1
strongly to the content and phenomenology of prospection. From an anatomical perspective, Institut des Neurosciences de la
Timone, Aix-Marseille University,
the visual system lies perfectly adjacent to posterior nodes of the core network, placing it in
27 Boulevard Jean Moulin,
a unique position to supply the visual elements necessary to simulate the future. Functional 13005 Marseille, France
2
neuroimaging studies of episodic future thinking consistently demonstrate increased activation The University of Sydney, Brain and
Mind Centre and School of Psychology,
and functional coupling between medial temporal regions and neocortical sites classically asso-
94 Mallett Street, Camperdown, NSW
ciated with visual imagery [5], even when object imagery performance is controlled for [6], pointing 2050, Australia
to a shared neural substrate underlying these capacities. Furthermore, behavioural studies
unequivocally reveal a foundational role for visual imagery during prospection [7], although current
approaches may be limited in their capacity to probe the visual contribution in all of its complexity
(Box 1). Finally, alterations in mental imagery capacity produce deleterious consequences for *Correspondence:
[email protected] (F. Conti)
future thinking and related constructive endeavours [8]. Accordingly, visual mental imagery and [email protected]
represents the primary sensory modality over which prospection operates, providing the essential (M. Irish).

272 Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 https://doi.org/10.1016/j.tics.2021.01.009
© 2021 Elsevier Ltd. All rights reserved.
Trends in Cognitive Sciences

Box 1. Representational Systems of the Brain

Glossary
Delineating the precise contribution of visual mental imagery to episodic future thinking requires us to disentangle its
Aphantasia: originally described by
recurrent association with another extensively invoked representational system (i.e., language). The tendency to situate
Francis Galton in 1880, the term was
words and images as orthogonal cognitive strategies to each other can be traced back to Ancient Greece, where Socrates
recently coined by A. Zeman to
affirmed that ‘the soul is ‘like a book’ into which ‘a writer within us’ inscribes ‘statements’ and ‘opinions’ (doxa). This writer,
describe a lack of mental imagery
however, does not work alone, because there is ‘a painter, who paints in our souls pictures (eikones) to illustrate the words
and an inability to voluntarily visualise
which the writer has written’ (reviewed in [88]). The emblematic metaphor of the painter and the scribe has led to the
imagery in the mind’s eye,
pervasive view that the mind operates via two fundamentally distinct representational systems [35]. Imagery, by its subjective
representing one extreme of a
nature, has proved particularly elusive to formal enquiry, whereas language has been regarded as the most accessible
putative imagery spectrum. The term
pathway to internal cognition. The ensuing language dominance has permeated through to the experimental approaches
derives from Greek ‘a-’ (i.e., ‘without’)
used to index prospection, whereby participants are typically asked to imagine future scenarios and describe the contents
and ‘phantasia’ (i.e., ‘imagination’).
of these simulations via verbal report. While intuitive, the reliability of this approach has been critiqued, given the
Core network: also known as the
requirement for participants to verbalise phenomenological characteristics such as vividness and sensory detail that would
default mode network, this network of
otherwise be inaccessible to the investigators [87]. The validity of future thinking tasks that rely on verbal report has further
brain regions is consistently implicated in
been questioned because the accuracy or veracity of the simulated event cannot be objectively determined. These
complex memory-/construction-based
criticisms aside, when the aforementioned measurements are used, it is difficult to tell whether task performance is
capacities, including remembering the
determined by language, imagery ability, or the capacity to successfully introspect on one’s internal milieu and to capture
past, imagining the future, and
the complexity of the scenario via verbal report. Studies of non-human animals or prelingual infants have demonstrated
constructing spatially contiguous
the need for alternative behavioural strategies that can reliably and efficiently index episodic content [89]. Comprehensive
scenes. Key regions in the core network
understanding of episodic future thinking requires experimental measures that allow us to access imagery content while
include the medial prefrontal cortex,
also being (i) comparable with objective standards and (ii) applicable to both clinical and healthy populations, as well as to
lateral temporal cortex, medial temporal
non-human species.
lobe (specifically the hippocampus),
retrosplenial cortex, posterior cingulate
cortex, precuneus, and inferior parietal
perceptual content and imbuing the resultant simulation with the phenomenological experience lobule.
of vividness. This integrative position enables us to account for the diversity of content and Gaze reinstatement: the spontaneous
phenomenology displayed across the healthy population during future thinking and provides a re-enactment of the eye movement
pattern produced during the initial
novel testbed to understand how alterations in visual imagery impact the constructive endeavour.
presentation of a visual stimulus.
Hyperphantasia: following the
Interplay between the Visual System and Core Network pioneering investigations of R.N. Haber,
From its location alone, the visuo-oculomotor brain system lies in an optimal position to influence this term describes individuals at the
other extreme of the putative imagery
constructive processes. This anatomical contiguity confers a massive adaptive advantage, spectrum, who present with extremely
enabling the seamless and bidirectional exchange of information between the visuo-oculomotor vivid and photo-like visual imagery.
system and the core network [9]. While the visual system supplies sensory content to the medial Simulation: mental recollection of a
past experience or the elaboration of a
temporal lobe (MTL), hippocampal activity has been shown to augment visual imagery
possible future scenario resulting from
processes and related oculomotor behaviour (e.g., execution of a saccade, gaze orientation) by the reinstantiation and integration of
influencing functional responses in cortical and subcortical regions of the oculomotor circuitry episodic and semantic representations.
[10]. Direct evidence for this bidirectional relationship was recently provided by Ryan et al. using Visual mental imagery: depictive
internal representation accompanied by
large-scale network dynamics to show that evoked responses in the hippocampus/MTL produce
the experience of sensory information in
observable responses in oculomotor areas within the time span of a typical gaze fixation [11]. the absence of a direct external stimulus.
Moreover, a recent human intracranial recording study in patients with epilepsy demonstrated It can be subdivided into at least two main
that hippocampal theta supports memory-guided eye movements during the encoding and re- components, known as spatial and object
visual mental imagery. Spatial imagery
trieval of novel object–location associations. Fixations to retrieved locations were time locked to
refers to the ability to elaborate abstract
the phase of the hippocampal theta rhythm, suggesting that the hippocampus supports representations of the spatial relationships
memory-guided exploration based on sampling prior experience and current perceptual states [12]. between objects, their locations, spatial
transformations, and movement. Object
imagery refers to the ability to construct
Further support for the functional and anatomical connections between the visuo-oculomotor pictorial images of items based on
and core network comes from the study of non-human primates [13,14]. Sharp wave ripples specific characteristics, such as, size,
(SWR) typically associated with memory consolidation during quiescent states have been shape, brightness, and colour.
observed during successful goal-directed visual exploration in the macaque [14], suggesting Vividness: phenomenological quality of
mental imagery indicating how well a
that hippocampal activity is tightly linked to patterns of oculomotor behaviour not only at rest representation can be visualised and
but also as an experience unfolds. Data-driven techniques using graph theoretic tools to derive experienced in the mind’s eye. Vividness
network-based algorithms underscore the natural emergence of distinct visuo-oculomotor and typically correlates with the subjective
feeling of re-experiencing a past memory
memory networks from the structural connectivity of the macaque [15]. Shen et al. [15] demon-
or pre-experiencing a future scenario.
strated that the neural pathways supporting the flow of information between the two emergent

Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 273

 Trends in Cognitive Sciences

systems converge on the dorsolateral prefrontal cortex (dlPFC) and anterior cingulate cortex
(ACC). These frontal regions were consistently, and somewhat unexpectedly, assigned to a
memory stream, given their strong connectivity with the hippocampus, despite being actively in-
volved in the cognitive control of saccades. A more detailed analysis of the average distance be-
tween nodes revealed that the majority of hippocampal subregions displayed disynaptic
connections to the frontal eye fields (FEFs) [15]. The Shen et al. study indicates that the FEF region
may play a pivotal role in directing information flow across the visuo-oculomotor and construction
systems, while cortical areas involved in oculomotor control, such as dlPFC and ACC, appear well
positioned to integrate this information to guide visual behaviour [15] (Figure 1).

Common Functional Properties of the Visual System and Core Network

Does the anatomical colocation of the visuo-oculomotor and core networks, and their reciprocal
flow of information, produce functionally relevant outcomes? Functional neuroimaging studies
independently exploring visual mental imagery [16] and future thinking [17] capacities consistently
converge on a distributed set of posterior parietal and occipital cortical regions, suggesting
common underlying functional properties. In the case of visual imagery (e.g., imagining a face
or place in response to a verbal cue), activation of posterior occipital and temporal regions, along-
side frontal and parietal regions, have been suggested to reflect the top-down rather than
bottom-up nature of this process [18]. The deliberate instantiation of mental imagery may reflect
the operations of a reverse hierarchy mechanism [19,20] initiated by the frontal cortex, which
triggers a cascade of activation that travels backwards in the cortical hierarchy, leading to
reactivation of sensory content from posterior cortices. While an attractive proposal, recent
empirical studies call into question this idea of a strict hierarchical division of labour, suggesting
that reactivation of visual features supporting memory processes is more widespread than
previously thought [21]. Nevertheless, this proposed cascade of activation during visual mental
imagery, whether hierarchical or not, bears strong similarities to prominent theories on the
dynamic reconstruction of episodic memories [22–24]. Similarly, studies of episodic future think-
ing demonstrate that this spread of cortical activation is accompanied by increased functional
connectivity between the hippocampus and regions of the visual cortex [25]. An important
point to clarify at this stage is that activation of visual areas during future thinking conditions is
not simply an artefact of object imagery. Addis et al. used a visual imagery control task whereby
participants imagined objects in a spatial arrangement, and they reported greater activation of the
core network during past and future event elaboration than the control task [6]. These findings
suggest that activation of visual areas during future thinking reflects their involvement in
processes beyond those implicated in object imagery.

Neural Mechanisms of Imagery-Driven Future Thinking

Future thinking may be endogenously generated or triggered by external stimuli, which are
primarily (but not always) visual in nature. Regardless of the trigger, theoretical accounts of mental
construction emphasise the ventromedial prefrontal cortex (vmPFC) as an important driver of
activity in the anterior hippocampus during mental construction [26]. The necessary spatial
backdrop of the simulation is assembled in the anterior hippocampus [27], anchored on a
conceptual semantic framework, which guides event construction according to a prescribed
semantic script/schema [28]. This scaffold is embellished with sensory-perceptual elements,
sourced from neocortical sites, and merged into a rich contextual layer via lateral posterior parietal
convergence zones (e.g., angular gyrus and posterior cingulate cortex) [29]. These contextually
rich assemblies are relayed to the hippocampus, where they can be reinstantiated as is in the
form of an episodic memory or flexibly recombined into novel future simulations [5]. The precise
pattern of neural activation in both visual and construction systems hinges upon the representa-
tional content invoked – the richer and more fine-grained the sensory-perceptual details, the

274 Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4

Trends in Cognitive Sciences

Trends in Cognitive Sciences

Figure 1. Process-Specific Alliances (PSAs) Supporting Visually Laden Expressions of Past and Future
Thinking. Three putative PSAs uniting visual mental imagery and oculomotor behaviour during past and future forms of
deliberate and spontaneous cognition. PSA 1 (shaded in green) supports the assembly of visual mental imagery into
contextually rich representations housed in posterior nodes of the core network, which are then relayed to the medial
temporal lobe (MTL) (in pink) to be assimilated into a spatially coherent event. A second PSA (PSA 2, in red) denotes a
coupling between the ventromedial prefrontal cortex (vmPFC) and the MTL in supporting schema-driven forms of event
construction. Finally, PSA 3 (shaded in blue) represents an oculomotor control process linking the cortical saccadic
network (blue) and the MTL (pink) in the consolidation and retrieval of episodic and semantic representations. Oculomotor
behaviour is guided via the integration of bottom-up and top-down goal-directed signals reflecting the contingent
characteristics of a given stimulus and any prior experience or knowledge of target identity and its semantic context,
respectively. Conversely, oculomotor variables such as gaze reinstatement actively support memory encoding and
retrieval, allowing the consolidation of recently acquired knowledge, as well as access to stored information on a moment-
to-moment basis. Brain template was obtained from Adobe Stock (https://stock.adobe.com) as a standard image under
license 84299969. Abbreviations: ACC, anterior cingulate cortex; dlPFC, dorsolateral prefrontal cortex; FEF, frontal eye
field; PCC, posterior cingulate cortex; RSC, retrosplenial cortex.

greater the posterior parietal and posterior hippocampal activation and corresponding functional
connectivity between the construction and visuo-oculomotor systems [30,31]. These changing
dynamics confer immense flexibility in how visual elements supplied by the visuo-oculomotor
system are co-opted by the construction system into a multidimensional mental experience
[32,33]. Moreover, the anatomical contiguity of these regions gives rise to an experience that is
predominantly visual in nature and experienced subjectively as vividness [34].

Individual Differences in Imagery Diverge across Past and Future Contexts

As with memory, imagery abilities vary considerably in the healthy population at large, prompting
calls for the reconceptualisation of imagery strength as a dimensional construct [35]. As such,
individuals can be situated along a continuum of imagery strength ranging from completely
absent (aphantasia) to extremely strong and photo-like (hyperphantasia) [36]. While the neural
basis of these imagery differences is unclear, a potential neural substrate has been proposed

Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 275

 Trends in Cognitive Sciences

whereby smaller visual cortices (V1 and V2) are associated with stronger but less precise visual
mental imagery in healthy individuals [37]. Individual differences in connectivity between
frontoparietal executive regions and sensory cortices representing modality-specific informa-
tion have further been suggested to influence the relative strength of imagery in the general
population [36].

This naturally occurring variation in imagery strength has important implications for how we recon-
struct the past and envisage the future [38]. Consistent with prominent theoretical accounts
emphasising the shared nature of memory and imagination [32], discrete aspects of visual mental
imagery underwrite many of the processes common to past and future thinking. The most
notable contribution is that of spatial imagery, concordant with prominent theories of scene
construction whereby construction of contextually rich 3D spatial imagery provides the requisite
backdrop for all forms of mental simulation [39]. Aydin reported that individual differences in
spatial imagery are the sole predictor of the level of episodic detail (i.e., event specificity) gener-
ated during past retrieval and future thinking [7], suggesting that scene imagery supports all
forms of mental simulation, regardless of temporal context or temporal distance [40]. At the
most impoverished extreme of the putative imagery continuum, aphantasic individuals who
experience an absence of internally generated mental imagery display parallel difficulties in
retrieving past events and imagining hypothetical scenarios in the future [8].

These commonalities aside, an individual’s proclivity for different imagery strategies (e.g., object
versus spatial) likely confers distinct benefits depending on the temporal demands of the
constructive task. For past retrieval, stronger object but not spatial imagery is associated with
higher levels of sensory detail and a greater sense of autonoetic re-experiencing [41]. The
Aydin study [7] further demonstrated an association between object imagery and discrete
aspects of past retrieval, including the number of visual details recalled, the overall coherence,
and the emotional intensity of the recalled experience, suggesting a central role in the
reinstantiation of sensory and emotional content experienced during event acquisition [24]. For
future thinking, however, object imagery predicted only the emotional intensity of simulated
events [7]. As such, individual differences in object imagery may diverge across past and future
contexts at the level of generated content versus subjective experience, respectively. Emerging
evidence suggests that spatial imagery is crucial for future simulation [42]. Individual differences
in visuospatial processing capacity correlate robustly with the provision of sensory-perceptual
details exclusively for future simulation, but not past retrieval [43]. Moreover, exposure to interfering
stimuli during memory retrieval and future thinking impacts discrete aspects of the constructive
endeavour depending on the individual’s underlying predisposition towards spatial or object
imagery. For example, when exposed to interfering stimuli, high spatial imagers are impeded in
the recollection and integration of spatial-contextual detail of a potential scenario to a greater extent
than high object imagers [41], whereas tasks requiring the reinstantiation of object imagery do not
appear susceptible to this disruption [44].

Age-Related Changes in Mental Imagery Compromise Prospection

Age-related changes in vision [45] and in visuospatial and visual exploration [46] processes
provide important convergent evidence regarding the synergies between visual imagery and
prospection. fMRI studies in healthy aging reveal a strong association between diminished
imagery strength and age-related memory impairment, attributable to reduced selectivity of visual
cortex activation during visual imagery, alongside decreased functional connectivity between
PFC and visual association cortices [47]. The reinstatement of cortical activity in posterior parietal
brain regions involved in the encoding of complex visual stimuli is further compromised in healthy
aging [48], disrupting the flow of information between these regions and other components of the

276 Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4

Trends in Cognitive Sciences

core network during the elaboration of vivid, imagery-rich scenarios. Mounting evidence from the
memory literature highlights a unique age-related disruption in the fidelity, specificity, and preci-
sion of contextual and item-specific information, even when age-related decreases in perceptual
sensitivity are considered [49]. Notably, age-related changes in constructive simulation and
imagery-related neural processes are accompanied by altered oculomotor behaviour, particularly
on tasks where visuospatial memory is required [46]. The evidence converges to suggest that
tasks which load heavily on visuospatial processes should be disproportionately affected in
older adults, as has been verified in studies of mental imagery [50] and future thinking [51]. An
intriguing observation is that the novelty of future simulations may be particularly vulnerable to
aging effects because older adults default to previously generated mental representations rather
than constructing novel future scenarios ex novo [52]. We suggest that age-related degradation
of stored visual percepts, and impaired access to these representations, not only impacts the
ability to recruit the content necessary for future simulations but also disrupts the ability to
discriminate between events that are highly similar or which share overlapping details [53]. By
this view, older adults are likely to recapitulate previously experienced events in their entirety
(i.e., ‘recasting’; see [54]) and to perform particularly poorly on experimental tasks that require
the extraction and flexible integration of multiple sensory and visuospatial elements [52].

Losing Sight of the Future

Compelling insights have been gleaned from studying alterations in future thinking capacity in
dementia populations, where underlying pathological processes result in a significant departure
from the cognitive trajectory of healthy aging [55]. To date, however, the study of mental imagery
in neurodegenerative disorders has proceeded largely independently from the literature on future
thinking with little to no crosstalk between these fields. The capacity for future thinking is particularly
vulnerable in Alzheimer’s disease, with the suggestion of a common pathological process driving
disturbances across past and future contexts [56]. While episodic memory has been proposed as
a potential mechanism, we suggest that degraded visual imagery processes offer a parsimonious
explanation for these impairments. Compromised retrieval of semantic information, coupled with
visuospatial and executive functioning deficits, has been shown to dramatically impair Alzheimer’s
patient performance on complex imagery tasks [57], with such impairments attributable to degener-
ation of posterior parietal brain regions [58]. On formal tasks of event construction, patients with
Alzheimer’s disease display a specific inability to access and harness visual elements to furnish
their mental constructions. For example, studies of autobiographical memory reveal a selective
decay of visual aspects of the recollective endeavour from early in the Alzheimer’s disease course,
with patients reporting that their retrieved memories more closely resemble static snapshots rather
than temporally extended events [59,60]. One study reported that >30% of autobiographical mem-
ories generated by patients with Alzheimer’s disease comprised tangential or non–self-referential im-
agery, whilst a further 16% of events were devoid of any visual imagery whatsoever [60].
Unsurprisingly, the capacity to mentally construct spatially contiguous scenes in the mind’s eye is
grossly impaired in Alzheimer’s disease, with patients rating their constructions as vague, divested
of rich perceptual detail, and typically re-experienced in black and white [61,62]. While the hippo-
campus might initially seem the most likely neural substrate of these disturbances, compelling
evidence points to degeneration of posterior parietal cortices in the genesis of early changes in
memory [63,64], episodic future thinking [65,66], and scene construction [61] in Alzheimer’s disease.
The assessment of these processes using narrative-based tasks in clinical populations, however, is
particularly challenging, suggesting the need for a new breed of objective measures.

Oculomotor Behaviour as a Tool to Understand Prospection

Gaze reinstatement provides a potentially powerful tool to explore foundational processes of
future thinking that are otherwise not amenable to self-report [46]. The reinstatement of eye

Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 277

 Trends in Cognitive Sciences

movements originally experienced during the encoding of an episode during retrieval bears
obvious parallels with cortical reinstatement effects observed on the neural level. A recent study
suggests that the hippocampus supports the reinstatement of highly specific event details
(e.g., people, locations) during future simulation and that this reinstatement covaries with the
vividness of the imagined event [67]. Memory studies incorporating eye tracking suggest that
participants spontaneously and unconsciously enact the original eye movement patterns
displayed during encoding to consolidate relational binding between items and the surrounding
context, facilitating access to previously formed memories about those items [68,69]. Gaze
reinstatement has been suggested to follow an inverted U-shaped curve during memory retrieval
[70], increasing when the mnemonic demands of the task are high [71], and the available
cognitive resources are scarce [72] up to a critical point beyond which gaze reinstatement may
no longer be necessary, or indeed available, to support performance. While an intriguing
proposal, further empirical data are required before we can reach any definitive conclusions
regarding this relationship.

Despite the considerable neural overlap between oculomotor behaviour and prospection, only a
handful of studies have investigated eye movements in the context of future thinking. De Vito et al.
proposed that while episodic memory and future thinking involve overlapping cognitive and neural
processes and are both expressed through complex mental imagery, their resultant outputs
comprise varying levels of spatial and object-based forms of mental imagery [73]. Participants
were required to conduct voluntary but guided eye movements during past and future thinking
(i.e., following a green dot as it moved horizontally across the screen). Execution of horizontal
saccades was found to significantly disrupt future thinking performance, with participants gener-
ating fewer episodic details relative to an unconstrained, freely moving condition [73]. In parallel,
participants generated an abundance of external (non-episodic) details, often taken to reflect off-
target or compensatory processes [74]. In this light, guided eye movements likely disrupt future
simulation by impeding spontaneous saccades, thereby precluding gaze reinstatement. By
compromising the curation and/or integration of relevant sensory-perceptual details into the
event simulation, guided eye movements may further precipitate a compensatory mechanism
whereby participants recruit accessible yet tangential external details to ‘fill in the blanks’ [75].

This interpretation fits well with a recent study that required healthy young participants to retrieve
past events and imagine possible future scenarios while looking at a blank screen [76]. Consistent
with previous reports, past experiences were rated as more vivid than future events and elicited
more fixations and saccades relative to future simulations. This oculomotor response was
interpreted by the authors as reflecting the specific operations of the visual system in the curation
(via saccades) and subsequent activation (via fixations) of the appropriate sensory-perceptual
content to populate the recollected mental scene [76]. This resonates with previous reports of
gaze fixations on the same regions of space as those inspected during initial stimulus presenta-
tion [77] and in change detection tasks where a greater number of fixations and saccades are
observed around the location of items that have been removed from the scene [78].

Despite these advances, it remains unclear to what extent spontaneous oculomotor behaviour
such as gaze reinstatement facilitates future thinking. This reflects the fact that the production
of eye movements itself has been suggested to increase cognitive load and competition for
available resources during future simulation [76]. In other words, the additional cognitive effort
imposed by the associative demands of future thinking might be predicted to produce a quanti-
tative reduction in eye movements. This would be most evident on open-ended laboratory-based
tasks where participants are instructed to envisage an unspecified event that will occur at some
point in the future (typically within the next 12 months). The ambiguity imposed by open-ended

278 Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4

Trends in Cognitive Sciences

tasks disproportionately taxes the semantic memory system, whereby participants must first
generate an appropriate semantic scaffold into which event details can be assimilated [33].
Under such conditions, the heightened constructive demands imposed by the open-ended
nature of the task might preclude spontaneous oculomotor behaviour. By contrast, on
well-defined tasks where the requisite semantic scaffold is embedded in task instructions
(e.g., scene construction tasks), greater incidence of spontaneous eye movements might be
predicted because the provision of the event cue considerably reduces the generative demands
of the task [33,79]. Recent studies incorporating pupillometry point to a positive correlation
between pupil dilation and retrieval/construction time during memory-based construction, most
prominent for future thinking [80], suggesting that pupil dilation may provide a useful marker of
future thinking from which the effect of cognitive load can be deduced [81].

Keeping an Eye on the Future – Towards an Integrative Framework

The extant literature converges to implicate interactions between the visuo-oculomotor and
hippocampal memory systems as central to visually laden forms of memory [70] (Box 2). Here,
we leverage these insights to suggest that specialised circuits converging on the hippocampus
mediate distinct aspects of future thinking, producing patterns of functional activity that converge
and diverge depending on task requirements. This idea resonates strongly with the proposal of
Process Specific Alliances (PSAs) in the brain, a ‘team’ of brain regions that rapidly assemble
to support a specific cognitive process yet quickly disassemble when the process is no longer
required [82]. As outlined by Cabeza and colleagues [82], PSAs offer a component-based
mechanistic account of specific neural operations performed in the service of an overarching
cognitive process. Where future thinking is concerned, PSAs represent an attractive candidate
mechanism – they are highly efficient and responsive to task demands, facilitating the flexible
reconfiguration of cortical activity in the face of changing task parameters [82].

Under this framework, we suggest that the hippocampus might form different PSAs with the core
memory network or visuo-oculomotor network to support different aspects of future thinking
depending on event novelty and task demands (Figure 1). For example, functional coupling
between the hippocampal/MTL and visuo-oculomotor systems during future thinking would

Box 2. Indexing Construction via Oculomotor Behaviour

To decipher how oculomotor behaviour might influence the construction of future events, it is first important to understand
the cognitive processes by which memory unfolds. In a seminal review, Ryan et al. provided an overview of how memory
and visual exploration likely intersect and the neurocognitive architecture underlying these processes [4]. First, the
execution of a saccade towards a specific target is proposed to be guided by the integration of bottom-up signals
reflecting the characteristics of a given stimulus (e.g., visual salience) with top-down goal-directed signals reflecting prior
experience, meaning, expectations, and knowledge of target identity and its semantic context [90,91]. The priority map
obtained from these sometimes competing signals is then instantiated by a network of regions involved in oculomotor
control [namely lateral intraparietal area (LIP), ACC, dlPFC, SEF, and FEF] to drive oculomotor behaviour [92].
Importantly, however, Ryan et al. [4] propose that this feedback loop linking the memory and oculomotor systems is
bidirectional in that eye movements support memory encoding, retrieval, and reconstruction processes as they unfold
[93]. Conversely, restricting visual exploration during encoding negatively affects subsequent retrieval [94]. Likewise,
maintaining central fixation during retrieval results in less detailed accounts, fewer spatiotemporal details, and slower
description times, likely because the cognitive and physiological resources typically required for event reconstruction are
co-opted to maintain fixation [95]. We suggest that these findings hint at a critical role for executive control in maintaining
gaze fixation (for a potential counterargument, see [96]), with activity in the dlPFC emerging as a possible neural substrate
for this process [97]. Finally, Liu et al. [98] demonstrated that neural activity along the ventral visual stream into the
hippocampus, as well as functional connectivity among the hippocampus, parahippocampal place area, and several other
cortical regions, is significantly reduced in fixed-viewing as opposed to free-viewing conditions. On the basis of these
results, Liu et al. suggested that oculomotor variables provide the informational units necessary for the formation of
new memories through hippocampal binding on a moment-to-moment basis [98] and therefore play a critical role in
supporting memory-/construction-based phenomena, including mind wandering and future thinking.

Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 279

 Trends in Cognitive Sciences

support the flexible extraction of sensory-perceptual representations from episodic memory,

whereas a second PSA involving the vmPFC and hippocampus might be instantiated when
schema-driven event construction is required [31]. A third candidate PSA in this context is the
partnership between the MTL and the cortical saccadic network (particularly FEF, dlPFC, and
ACC) in the control of eye movements, because many of these regions are well positioned to
guide the relay of information between the memory and the visuo-oculomotor system during ep-
isodic simulation processes [15]. In our view, the activation of cortical regions involved in oculo-
motor control is triggered by the dlPFC. Oculomotor behaviour is guided by previously
established knowledge, and, conversely, the recollection of previously stored information or epi-
sodic detail is facilitated by visual exploration. As such, we propose that information flow between
the visuo-oculomotor and hippocampal/MTL systems is channelled by disynaptic connections
converging on FEF, whereby the quantity and quality of evoked oculomotor responses is ex-
pected to reflect the novelty and level of complexity of the simulation.

While the capacity to rapidly assemble and reconfigure PSAs is predicted to confer immense
flexibility during future simulation, the time course of such dynamics remains unclear. The
repeated simulation of future events has been shown to yield event constructions that progres-
sively resemble episodic memories in terms of their construction times and incorporated detail
[42]. As such, it may be possible to use repetition paradigms to determine the various stages
of coupling and uncoupling between the proposed PSAs. Studies leveraging the high temporal
resolution of magnetoencephalography (MEG) may prove particularly useful in this regard,
enabling us to determine the time course of functional reorganisation between putative PSAs
bridging the visual system and core network, as well as elucidating how such dynamics fluctuate
across past and future temporal contexts.

Building on the extant episodic memory literature, we further propose that the oculomotor
response elicited during future thinking follows an inverted U-shaped curve. Eye movements
are predicted to scale with task complexity up to a critical threshold beyond which attentional
resources must be reallocated to cope with increasing task demands, producing an attenuation
of overt eye movements. Given the heightened constructive demands when events are imagined
for the first time, and given that future simulations become increasingly similar to past events
across repetitions [42], an intuitive proposal is that the eye movement patterns drawn during
the encoding of a memory or the initial construction of a future scenario are reinstated with
subsequent retrieval or repeated simulation. Truly novel and elaborate spatial contexts might
induce less pronounced eye movement patterns than more predictable, simple, or similar to
already encountered spatial contexts that can be recapitulated in full, or partially sampled, from
long-term memory.

However, it remains unclear what profile of eye movements to expect when individuals envis-
age the future during periods of minimal cognitive demand (e.g., mind wandering). Although
we might be tempted to predict an increase in eye movements due to the spontaneous and
minimally demanding nature of this endeavour, the perceptual decoupling from the external
environment that is characteristic of mind wandering [83] might produce a dampening of
conspicuous oculomotor behaviour altogether [84,85]. Studies decoding periods of mind
wandering from the eye movements of healthy participants during reading [84] suggest that
reduced processing of the external environment during periods of internally focused attention
may be objectively measured using oculomotor measures [85]. Articulating the conditions by
which oculomotor behaviour scales relative to the changing cognitive demands imposed by
spontaneous versus deliberate expressions of future thinking will be an important question
to address.

280 Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4

Trends in Cognitive Sciences

Box 3. Insights from Clinical Populations Outstanding Questions

How might we use these emergent findings to better understand the symptomatology and cognitive sequelae of clinical
Mental imagery is inherently multifaceted.
disorders? Reductions in the size and total number of neurons in V1 have been observed in clinical populations such as
Are there fundamental differences in the
schizophrenia and post-traumatic stress disorder, and exceptionally vivid imagery is a recurrent feature of these disorders
representational format of other types of
[94,95]. One clinical population of immense interest in this context is Parkinson’s disease. Visual disturbances, including
mental imagery (e.g., motor, auditory)?
visual hallucinations, are prominent nonmotor symptoms in Parkinson’s disease, although the underlying mechanisms
How might we objectively query mental
remain unclear [99]. Similarly, an impaired capacity for prospection has been reported in this syndrome in the absence
imagery in all its manifestations to better
of overt changes in episodic memory integrity [100]. A recent study exploring the related phenomenon of spontaneous
understand the phenomenology of future
cognition in Parkinson’s disease offers important clues for future study [101]. Patients with Parkinson’s disease who
thinking?
experience visual hallucinations were found to display an increased propensity for mind wandering relative to their
nonhallucinating counterparts. Inter-network connectivity and seed-to-voxel analyses indicated that increased mind
How does temporal distance from the
wandering in hallucinating versus nonhallucinating patients reflected stronger coupling between the primary visual cortex
current moment influence the vividness
and dorsal portions of the core network, offering a novel neural substrate for aberrant internally generated visual
of instantiated visual imagery when
content [101]. Observation of increased visual system–core network coupling provides an intriguing foundation to unravel
envisaging the distant future? Could we
top-down influences over perception and their impact on deliberate (e.g., prospection) as well as involuntary
train people to improve their imagery
(e.g., hallucinations) manifestations of visual imagery. An open question in this regard is whether visually guided behaviour
strength in order to foster more prudent
can be harnessed to objectively quantify atypical oculomotor signatures of aberrant mental construction processes. When
decision-making?
used in combination with computational modelling and machine learning tools, eye-tracking paradigms yield distinct
behavioural biomarkers to accurately classify developmental disorders such as autism and attention-deficit/hyperactivity Is imagery strength the key to envisaging
disorder (ADHD) [102]. Moreover, studies of oculomotor behaviour in older adults demonstrate the utility of eye move- the future, or is it more a case of imagery
ments as a potentially useful tool in the early detection of neurodegeneration [103]. While a promising start, we suggest control?
that a convergent approach will be essential to elucidate how alterations in visual imagery give rise to maladaptive forms
of future thinking and spontaneous cognition, providing a powerful framework to track the unfolding of these processes Is prospection possible in the absence
in health and disease. of visual imagery?

How does emotion influence mental

Finally, clinical populations with reduced cognitive resources for whom introspection regarding imagery, and, in turn, how does this
impact the representational quality of
internal mental states is particularly challenging [86] might be expected to show diminished
our future simulations?
oculomotor behaviour during future thinking tasks. Whether it is possible to engage in prospection
or truly episodic recollection in the absence of visual mental imagery represents an intriguing Are there cultural differences in how
question. The recent study on congenital aphantasia by Dawes et al. suggests that lower-level or mental imagery is invoked when
constructing the future, and how
rudimentary forms of memory and future thinking may be possible, although the level of episodic
might we objectively measure such
detail and perceived vividness of future simulations is markedly diminished [8]. Translating these differences?
insights to neurological disorders has the potential to refine our understanding of graded variations
in the phenomenology of past and future thinking in health and disease (Box 3). Do animals possess mental imagery?
Can we translate visuo-oculomotor
paradigms to better understand the in-
Concluding Remarks ternal milieu of non-human primates?
We routinely ‘look ahead,’ ‘envisage,’ or ‘see ourselves’ at future time points. Such expressions
How does intentionality impact the
reflect the inherently visual nature of prospection and the tight coupling between visual imagery
vividness and imagery strength of
and event construction in general. While the language we use seems to accurately capture our future simulations? Do spontaneous
proclivity for prospection, humans appear far less adept at introspecting and reporting reliably or involuntary forms of mental imagery
upon the visual contents of their internal milieu [87]. This is compounded in clinical and develop- differ substantially from deliberate or
voluntary forms? Could we leverage
mental populations where the contents and phenomenology of prospection may not be amenable
such differences to better understand
to verbal report, suggesting the need for objective assays to tap these elusive processes (see maladaptive expressions of visual
Outstanding Questions). Oculomotor variables extracted from eye-tracking paradigms represent imagery such as hallucinations?
one such promising avenue for further research, enabling us to derive surrogate markers of past
Do current scoring protocols for past-
and future thinking and their respective phenomenology. As the field moves progressively towards and future-oriented construction ade-
more nuanced and integrative positions, we envision that inclusion of objective indices of this quately capture the imagery component?
nature will help to reposition visual mental imagery as an integral cog in the simulation machine. How can we optimise the way in which
we extract sensory-perceptual informa-
tion from participant narratives to better
Acknowledgments understand the interplay between visual
This work was supported, in part, by a Jean-Walter Zellidja Fellowship to F.C. provided by the Académie Française for the mental imagery and event construction?
academic year 2019/2020, as well as by the A*Midex Foundation and the French National Research Agency funded by the
French government Investissements d’Avenir programme (NeuroSchool, nEURo*AMU, ANR-17-EURE-0029 grant). M.I. is
supported by an Australian Research Council Future Fellowship (FT160100096) and an Australian Research Council
Discovery Project (DP180101548).

Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 281

 Trends in Cognitive Sciences

Declaration of Interests
The authors report no conflict of interest.

References
1. Bulley, A. and Irish, M. (2018) The functions of prospection – 27. Zeidman, P. and Maguire, E.A. (2016) Anterior hippocampus:
variations in health and disease. Front. Psychol. 9, 2328 the anatomy of perception, imagination and episodic memory.
2. Schacter, D.L. et al. (2017) Episodic future thinking: mechanisms Nat. Rev. Neurosci. 17, 173–182
and functions. Curr. Opin. Behav. Sci. 17, 41–50 28. Gilboa, A. and Marlatte, H. (2017) Neurobiology of schemas
3. Aristotle (1984) De Anima, Princeton University Press and schema-mediated memory. Trends Cogn. Sci. 21,
4. Ryan, J.D. et al. (2020) The intersection between the 618–631
oculomotor and hippocampal memory systems: empirical 29. Ramanan, S. et al. (2018) Rethinking the role of the angular
developments and clinical implications. Ann. N. Y. Acad. gyrus in remembering the past and imagining the future – the
Sci. 1464, 115–141 Contextual Integration Model. Neuroscientist 24, 342–352
5. Schacter, D.L. et al. (2012) The future of memory: remembering, 30. Brunec, I.K. et al. (2018) Multiple scales of representation along
imagining, and the brain. Neuron 76, 677–694 the hippocampal anteroposterior axis in humans. Curr. Biol.
6. Addis, D.R. et al. (2007) Remembering the past and imagining 28, 2129–2135
the future: common and distinct neural substrates during 31. Irish, M. and Vatansever, D. (2020) Rethinking the episodic-
event construction and elaboration. Neuropsychologia 45, semantic distinction from a gradient perspective. Curr. Opin.
1363–1377 Behav. Sci. 32, 43–49
7. Aydin, C. (2018) The differential contributions of visual 32. Addis, D.R. (2018) Are episodic memories special? On the
imagery constructs on autobiographical thinking. Memory sameness of remembered and imagined event simulation.
26, 189–200 J. R. Soc. N. Z. 48, 64–88
8. Dawes, A.J. et al. (2020) A cognitive profile of multi-sensory 33. Irish, M. (2020) On the interaction between episodic and
imagery, memory and dreaming in aphantasia. Sci. Rep. 10, semantic representations - constructing a unified account of
10022 imagination. In The Cambridge Handbook of the Imagination
9. Rosen, M.L. et al. (2018) The role of visual association cortex in (Abraham, A., ed.), pp. 447–465, Cambridge University Press
associative memory formation across development. J. Cogn. 34. Richter, F.R. et al. (2016) Distinct neural mechanisms underlie
Neurosci. 30, 365–380 the success, precision, and vividness of episodic memory.
10. Parr, T. and Friston, K.J. (2017) The active construction of the eLife 5, e18260
visual world. Neuropsychologia 104, 92–101 35. MacKisack, M. (2018) Painter and scribe: From model of mind
11. Ryan, J.D. et al. (2020) Modeling the influence of the to cognitive strategy. Cortex 105, 118–124
hippocampal memory system on the oculomotor system. 36. Zeman, A. et al. (2020) Phantasia – the psychological signifi-
Netw. Neurosci. 4, 217–233 cance of lifelong visual imagery vividness extremes. Cortex
12. Kragel, J.E. et al. (2020) Hippocampal theta coordinates memory 130, 426–440
processing during visual exploration. Elife 9, e52108 37. Bergmann, J. et al. (2016) Smaller primary visual cortex is
13. Jutras, M.J. et al. (2013) Oscillatory activity in the monkey associated with stronger, but less precise mental imagery.
hippocampus during visual exploration and memory formation. Cereb. Cortex 26, 3838–3850
Proc. Natl. Acad. Sci. U. S. A. 110, 13144–13149 38. D’Argembeau, A. and Van der Linden, M. (2006) Individual
14. Leonard, T.K. et al. (2015) Sharp wave ripples during visual differences in the phenomenology of mental time travel: the
exploration in the primate hippocampus. J. Neurosci. 35, effect of vivid visual imagery and emotion regulation strategies.
14771–14782 Conscious. Cogn. 15, 342–350
15. Shen, K. et al. (2016) An anatomical interface between memory 39. Mullally, S.L. and Maguire, E.A. (2014) Memory, imagination,
and oculomotor systems. J. Cogn. Neurosci. 28, 1772–1783 and predicting the future: a common brain mechanism?
16. Winlove, C.I.P. et al. (2018) The neural correlates of visual Neuroscientist 20, 220–234
imagery: a co-ordinate-based meta-analysis. Cortex 105, 4–25 40. Barry, D.N. and Maguire, E.A. (2019) Remote memory and the
17. Beaty, R.E. et al. (2018) Core network contributions to remem- hippocampus: a constructive critique. Trends Cogn. Sci. 23,
bering the past, imagining the future, and thinking creatively. 128–142
J. Cogn. Neurosci. 30, 1939–1951 41. Sheldon, S. et al. (2017) Individual differences in visual imagery
18. Pearson, J. (2019) The human imagination: the cognitive determine how event information is remembered. Memory 25,
neuroscience of visual mental imagery. Nat. Rev. Neurosci. 360–369
20, 624–634 42. Wiebels, K. et al. (2020) Relational processing demands and the
19. Dijkstra, N. et al. (2017) Distinct top-down and bottom-up brain role of spatial context in the construction of episodic simulations.
connectivity during visual perception and imagery. Sci. Rep. 7, J. Exp. Psychol. Learn. Mem. Cogn. 46, 1424–1441
5677 43. D’Argembeau, A. et al. (2010) Component processes underlying
20. Dentico, D. et al. (2014) Reversal of cortical information flow future thinking. Mem. Cogn. 38, 809–819
during visual imagery as compared to visual perception. 44. de Vito, S. et al. (2014) Eye movements disrupt spatial but not
Neuroimage 100, 237–243 visual mental imagery. Cogn. Process. 15, 543–549
21. Bone, M.B. et al. (2020) Feature-specific neural reactivation 45. Owsley, C. (2011) Aging and vision. Vis. Res. 51, 1610–1622
during episodic memory. Nat. Commun. 11, 1945 46. Wynn, J.S. et al. (2020) How older adults remember the
22. Ranganath, C. and Ritchey, M. (2012) Two cortical systems for world depends on how they see it. Trends Cogn. Sci. 24,
memory-guided behaviour. Nat. Rev. Neurosci. 13, 713–726 858–861
23. Renoult, L. et al. (2019) From knowing to remembering: 47. Kalkstein, J. et al. (2011) Diminished top-down control
the semantic-episodic distinction. Trends Cogn. Sci. 23, underlies a visual imagery deficit in normal aging. J. Neurosci.
1041–1057 31, 15768–15774
24. Danker, J.F. and Anderson, J.R. (2010) The ghosts of brain 48. Folville, A. et al. (2020) Age-related differences in the neural
states past: remembering reactivates the brain regions correlates of vivid remembering. Neuroimage 206, 116336
engaged during encoding. Psychol. Bull. 136, 87–102 49. Korkki, S.M. et al. (2020) Healthy ageing reduces the precision
25. Bellana, B. et al. (2017) Similarities and differences in the of episodic memory retrieval. Psychol. Aging 35, 124–142
default mode network across rest, retrieval, and future imagining. 50. Castellano, S. et al. (2015) Imagery in healthy and in cognitively
Hum. Brain Mapp. 38, 1155–1171 impaired aging. Clin. Gerontol. 38, 103–113
26. Monk, A.M. et al. (2021) The role of hippocampal-ventromedial 51. Gaesser, B. et al. (2011) Characterizing age-related changes in
prefrontal cortex neural dynamics in building mental representations. remembering the past and imagining the future. Psychol. Aging
J. Cogn. Neurosci. 33, 89–103 26, 80–84

282 Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4

Trends in Cognitive Sciences

52. Schacter, D.L. et al. (2013) Remembering the past and imagining 77. Laeng, B. et al. (2014) Scrutinizing visual images: the role of
the future in the elderly. Gerontology 59, 143–151 gaze in mental imagery and memory. Cognition 131, 263–283
53. Trelle, A.N. et al. (2019) Neural evidence for age-related differences 78. Scholz, A. et al. (2015) Eye movements reveal memory
in representational quality and strategic retrieval processes. processes during similarity- and rule-based decision making.
Neurobiol. Aging 84, 50–60 Cognition 136, 228–246
54. Irish, M. (2016) Semantic memory as the essential scaffold 79. Sheldon, S. and Levine, B. (2016) The role of the hippocampus
for future oriented mental time travel. In Seeing the Future: in memory and mental construction. Ann. N. Y. Acad. Sci.
Theoretical Perspectives on Future-Oriented Mental Time 1369, 76–92
Travel (Michaelian, K. et al., eds), pp. 388–408, 80. El Haj, M. and Moustafa, A.A. (2020) Pupil dilation as an indica-
Oxford University Press tor of future thinking. Neurol. Sci. Published online July 10,
55. Andrews-Hanna, J.R. et al. (2019) A review and reappraisal of 2020. https://doi.org/10.1007/s10072-020-04533-z
the default network in normal aging and dementia. In Oxford En- 81. Wahn, B. et al. (2016) Pupil sizes scale with attentional load
cyclopedia of Psychology and Aging (Knight, B.G., ed.), Oxford and task experience in a multiple object tracking task. PLoS
University Press One 11, e0168087
56. Addis, D.R. et al. (2009) Episodic simulation of future events is 82. Cabeza, R. et al. (2018) Process-specific alliances (PSAs) in
impaired in mild Alzheimer’s disease. Neuropsychologia 47, cognitive neuroscience. Trends Cogn. Sci. 22, 996–1010
2660–2671 83. Smallwood, J. et al. (2008) Going AWOL in the brain: mind
57. Hussey, E.P. et al. (2012) Using mental imagery to improve wandering reduces cortical analysis of external events.
memory in patients with Alzheimer disease: trouble generating J. Cogn. Neurosci. 20, 458–469
or remembering the mind’s eye? Alzheimer Dis. Assoc. Disord. 84. Reichle, E.D. et al. (2010) Eye movements during mindless
26, 124–134 reading. Psychol. Sci. 21, 1300–1310
58. Salimi, S. et al. (2018) Can visuospatial measures improve 85. Salvi, C. and Bowden, E.M. (2016) Looking for creativity:
the diagnosis of Alzheimer’s disease? Alzheimers Dement. where do we look when we look for new ideas? Front. Psychol.
(Amst.) 10, 66–74 7, 161
59. Irish, M. et al. (2010) Exploring the recollective experience during 86. O’Callaghan, C. et al. (2019) Hippocampal atrophy and intrinsic
autobiographical memory retrieval in amnestic mild cognitive brain network dysfunction relate to alterations in mind wandering
impairment. J. Int. Neuropsychol. Soc. 16, 546–555 in neurodegeneration. Proc. Natl. Acad. Sci. U. S. A. 116,
60. Irish, M. et al. (2011) Impaired capacity for autonoetic reliving 3316–3321
during autobiographical event recall in mild Alzheimer’s 87. Miloyan, B. and McFarlane, K.A. (2019) The measurement of
disease. Cortex 47, 236–249 episodic foresight: a systematic review of assessment instruments.
61. Irish, M. et al. (2015) Scene construction impairments in Cortex 117, 351–370
Alzheimer’s disease – a unique role for the posterior cingulate 88. MacKisack, M. et al. (2016) On picturing a candle: the prehis-
cortex. Cortex 73, 10–23 tory of imagery science. Front. Psychol. 7, 515
62. Wilson, N.A. et al. (2020) Scene construction impairments in 89. Redshaw, J. and Bulley, A. (2018) Future thinking in animals –
frontotemporal dementia: evidence for a primary hippocampal capacities and limits. In The Psychology of Thinking About the
contribution. Neuropsychologia 137, 107327 Future (Oettingen, G. et al., eds), pp. 31–51, Guilford Press
63. Irish, M. et al. (2014) Common and unique gray matter corre- 90. Draschkow, D. et al. (2014) Seek and you shall remember:
lates of episodic memory dysfunction in frontotemporal scene semantics interact with visual search to build better
dementia and Alzheimer’s disease. Hum. Brain Mapp. 35, memories. J. Vis. 14, 10
1422–1435 91. Henderson, J.M. and Hayes, T.R. (2017) Meaning-based
64. Irish, M. et al. (2014) Grey and white matter correlates of recent guidance of attention in scenes as revealed by meaning maps.
and remote autobiographical memory retrieval – insights from Nat. Hum. Behav. 1, 743–747
the dementias. PLoS One 9, e113081 92. Bisley, J.W. and Mirpour, K. (2019) The neural instantiation of a
65. Irish, M. et al. (2012) Considering the role of semantic memory priority map. Curr. Opin. Psychol. 29, 108–112
in episodic future thinking: evidence from semantic dementia. 93. Ryan, J.D. and Shen, K. (2019) The eyes are a window into
Brain 135, 2178–2191 memory. Curr. Opin. Behav. Sci. 32, 1–6
66. Irish, M. et al. (2013) Episodic future thinking is impaired in the 94. Damiano, C. and Walther, D.B. (2019) Distinct roles of eye
behavioural variant of frontotemporal dementia. Cortex 49, movements during memory encoding and retrieval. Cognition
2377–2388 184, 119–129
67. Thakral, P.P. et al. (2020) Reinstatement of event details during 95. Lenoble, Q. et al. (2019) Don’t stare, unless you don’t want to
episodic simulation in the hippocampus. Cereb. Cortex 30, remember: maintaining fixation compromises autobiographical
2321–2337 memory retrieval. Memory 27, 231–238
68. Wynn, J.S. et al. (2016) Selective scanpath repetition during 96. Armson, M.J. et al. (2019) Maintaining fixation does not
memory-guided visual search. Vis. Cogn. 24, 15–37 increase demands on working memory relative to free viewing.
69. Wynn, J.S. et al. (2018) Fixation reinstatement supports visuo- PeerJ 7, e6839
spatial memory in older adults. J. Exp. Psychol. Hum. Percept. 97. Henderson, J.M. and Choi, W. (2015) Neural correlates of fixa-
Perform. 44, 1119–1127 tion duration during real-world scene viewing: evidence from
70. Wynn, J.S. et al. (2019) Eye movements actively reinstate fixation-related (FIRE) fMRI. J. Cogn. Neurosci. 27, 1137–1145
spatiotemporal mnemonic content. Vision (Basel) 3, 21 98. Liu, Z.X. et al. (2020) Restricting visual exploration directly
71. Olsen, R.K. et al. (2014) The relationship between delay period impedes neural activity, functional connectivity, and memory.
eye movements and visuospatial memory. J. Vis. 14, 8 Cereb. Cortex Commun. 1, tgaa54
72. Grady, C. (2012) The cognitive neuroscience of ageing. Nat. 99. Weil, R.S. et al. (2016) Visual dysfunction in Parkinson’s
Rev. Neurosci. 13, 491–505 disease. Brain 139, 2827–2843
73. de Vito, S. et al. (2015) Eye movements disrupt episodic future 100. de Vito, S. et al. (2012) Future thinking in Parkinson’s disease:
thinking. Memory 23, 796–805 an executive function? Neuropsychologia 50, 1494–1501
74. Strikwerda-Brown, C. et al. (2019) External details revisited – a 101. Walpola, I.C. et al. (2020) Mind wandering in Parkinson’s
new taxonomy for coding ‘non-episodic’ content during auto- disease hallucinations reflects primary visual and default
biographical memory retrieval. J. Neuropsychol. 13, 371–397 network coupling. Cortex 125, 233–245
75. Irish, M. et al. (2012) Exploring the content and quality of episodic 102. Itti, L. (2015) New eye-tracking techniques may revolutionize
future simulations in semantic dementia. Neuropsychologia 50, mental health screening. Neuron 88, 442–444
3488–3495 103. Whitehead, J.C. et al. (2018) Portable eyetracking-based
76. El Haj, M. and Lenoble, Q. (2018) Eying the future: eye assessment of memory decline. J. Clin. Exp. Neuropsychol.
movement in past and future thinking. Cortex 105, 97–103 40, 904–916

Trends in Cognitive Sciences, April 2021, Vol. 25, No. 4 283