In recent years, scientists have increasingly taken to investigate the predictive nature of cognition. We argue that prediction relies on abstraction, and thus theories of predictive cognition need an explicit theory of abstract representation. We propose such a theory of the abstract representational capacities that allow humans to transcend the “here-and-now.” Consistent with the predictive cognition literature, we suggest that the representational substrates of the mind are built as a hierarchy, ranging from the concrete to the abstract; however, we argue that there are qualitative differences between elements along this hierarchy, generating meaningful, often unacknowledged, diversity. Echoing views from philosophy, we suggest that the representational hierarchy can be parsed into: modality-specific representations, instantiated on perceptual similarity; multimodal representations, instantiated primarily on the discovery of spatiotemporal contiguity; and categorical representations, instantiated primarily on social interaction. These elements serve as the building blocks of complex structures discussed in cognitive psychology (e.g., episodes, scripts) and are the inputs for mental representations that behave like functions, typically discussed in linguistics (i.e., predicators). We support our argument for representational diversity by explaining how the elements in our ontology are all required to account for humans’ predictive cognition (e.g., in subserving logic-based prediction; in optimizing the trade-off between accurate and detailed predictions) and by examining how the neuroscientific evidence coheres with our account. In doing so, we provide a testable model of the neural bases of conceptual cognition and highlight several important implications to research on self-projection, reinforcement learning, and predictive-processing models of psychopathology.

Autism has been described as a neural deficit in prediction, people with autism manifest low perceptual construal and are impaired at traversing psychological distances, and Gilead et al.'s hierarchy from iconic to multimodal to fully abstract, socially communicated representations is exactly the hierarchy of representational impairment in autism, making autism a natural behavioural and neurophysiological test case for the prediction–abstraction relationship.

Thinking about what the senses cannot grasp is one of the hallmarks of human cognition. We argue that “intangible abstracta” are represented differently from other products of abstraction, that goal-derived categorization supports their learning, and that they are grounded also in internalized linguistic and social interaction. We conclude by suggesting different ways in which abstractness contributes to cement group cohesion.

The authors’ arguments reflect the dominant traditions of American Psychology. In doing so, however, they miss relevant insights omitted during the original importation (translation and popularization) of the foreign sources that informed the theories they built upon. Of particular relevance here are Piaget's last studies. These are presented to unpack the meaning of “object permanence” as a kind of representation.

Gilead et al.'s approach to human cognition places abstraction and prediction at the heart of “mental travel” under a “representational diversity” perspective that embraces foundational concepts in cognitive science. But, it gives insufficient credit to the possibility that the process of abstraction produces a gradient, and underestimates the importance of a highly influential domain in predictive cognition: language, and related, the emergence of experientially based structure through time.

Bayesian decision theory provides a simple formal elucidation of some of the ways that representation and representational abstraction are involved with, and exploit, both prediction and its rather distant cousin, predictive coding. Both model-free and model-based methods are involved.

We focus on the theory of abstraction proposed by the target article. We suggest that abstraction varies at different levels of learning, cognitive development, or cognitive ability. We argue that this theory does not specify how abstraction is done at each of these levels. Because of these weaknesses, the theory cannot explicate how individuals differ in mental time travel at different phases of life or different levels of cognitive ability.

If we consider perceptions as arising from predictive processes, we must consider the manner in which the underlying expectations are formed and how they are applied to the sensory data. We provide examples of cases where expectations give rise to unexpected and unlikely perceptions of the world. These examples may help define bounds for the notion that perceptual hypotheses are direct derivatives of experience and are used to furnish sensible interpretations of sensory data.

Prediction draws on both simulation and theory. I ask how simulation is defined, and what the roles of simulation and theory are, respectively. Simulation is flexible in structure and resources. Often simulation and theory are combined in prediction. The function of simulation consists of representing a situation that is relevantly like the target situation with regards to the feature predicted.

Gilead et al. state that abstraction supports mental travel, and that mental travel critically relies on abstraction. I propose an important addition to this theoretical framework, namely that mental travel might also support abstraction. Specifically, I argue that spontaneous mental travel (mind wandering), much like data augmentation in machine learning, provides variability in mental content and context necessary for abstraction.

Explanations of how the brain makes successful predictions should refer to abstracta. But, the mind/brain system is for more than prediction alone. Creativity also plays an important role in supply the mind/brain system with abstracta that serve a number of valuable ends over and above prediction.

In their scholarly target article, Gilead et al. explain how abstract mental representations and the predictive brain enable prospection and time-traveling. However, their exclusive focus on intrapsychic capacities misses an important point, namely, the degree to which mind and brain are tuned by the environment. This neglected aspect of adaptive cognition is discussed and illustrated from a cognitive-ecological perspective.

Gilead et al. propose an ontology of abstract representations based on folk-psychological conceptions of cognitive architecture. There is, however, no evidence that the experience of cognition reveals the architecture of cognition. Scale-free architectural models propose that cognition has the same computational architecture from sub-cellular to whole-organism scales. This scale-free architecture supports representations with diverse functions and levels of abstraction.

Gilead et al. raise some fascinating issues about representational substrates and structures in the predictive brain. This commentary drills down on a core theme in their arguments; namely, the structure of models that generate predictions. In particular, it highlights their factorial nature – both in terms of deep hierarchies over levels of abstraction and, crucially, time – and how this underwrites agency.

Instead of endorsing an all-encompassing view about the influence of abstractions in predictive processing, I suggest that most deliberative thought including complex abstractions, agent actions, and/or perceived environmental sequences are stored in the human prefrontal cortex in the form of structured event complexes.

We leverage the notion that abstraction enables prediction to generate novel insights and hypotheses for the literatures on attitudes and mate preferences. We suggest that ideas about liking (e.g., evaluations of categories or overall traits) are more abstract than experiences of liking (e.g., evaluations of particular exemplars), and that ideas about liking may facilitate mental travel beyond the here-and-now.

Predictive processing models of psychopathologies are not explanatorily consistent with the present account of abstract thought. These models are based on latent variables probabilistically mapping the structure of the world. As such, they cannot be informed by representational ontology based on mental objects and states. What actually is the case is merely some terminological affinity between subjective and informational uncertainty.

A crucial aspect of Gilead and colleagues’ ontology is the dichotomy between tangible and intangible representations, but the latter remains rather ill-defined. We propose a fundamental role for interoceptive experience and the statistical distribution of entities in language, especially for intangible representations, that we believe Gilead and colleagues’ ontology needs to incorporate.

Gilead et al. present a rich account of abstraction. Though the account describes several elements which influence mental representation, it is worth also delineating how feelings, such as fluency and emotion, influence mental simulation. Additionally, though past experience can sometimes make simulations more accurate and worthwhile (as Gilead et al. suggest), many systematic prediction errors persist despite substantial experience.

Gilead et al.'s theory presupposes that traversing temporal, spatial, social, and hypothetical distances are largely interchangeable acts of mental travel that co-occur in human ontogeny. Yet, this claim is at odds with recent developmental data suggesting that children's reasoning is differentially affected by the dimension which they must traverse, and that different representational abilities underlie travel across different dimensions.