Over the two decades, a growing body of evidence has revealed the previously unanticipated complexity of insect behavior and cognition (Döring and Chittka, 2011). Studies focusing predominantly on species such as bees (Giurfa, 2007), wasps (Weise et al., 2022), ants (Czaczkes, 2022), but also fruit flies (Grover et al., 2022), crickets (Baran et al., 2023), and locusts (Ben-Nun et al., 2013), have demonstrated a broad range of complex capacities, including associative and observational learning, decision-making, rule generalization, and even the abstraction of symbolic information. Such findings challenge long-standing views that insect behavior is predominantly reflexive, rigid, and governed solely by stimulus-response mechanisms (Giurfa, 2015). Insects, despite their comparatively diminutive nervous systems, exhibit cognitive and behavioral capacities that seem to parallel those observed in animals with larger brains (Perry et al., 2017).

This raises a foundational question: How should these capacities be accounted for? Are they a product of internal control systems akin to those supporting higher cognition in vertebrates, or do they emerge from highly domain-specific associative learning or from sets of heuristics precisely fine-tuned over evolutionary history? Current interpretations are often polarized. Those inclined toward reductive explanations point out the danger of falling for undue anthropomorphic projections (Dhein, 2023). At the same time, their opponents raise the issue of a lack of evolutionary justification for excessive reductionism (Hohol et al., 2017a). Indeed, insect neuroanatomy and neurophysiology is immensely dissimilar to the vertebrate one, yet conversely, in theory, arbitrary complex behavior may be reduced to a sufficiently extensive set of simpler actions (Skinner, 1938). Nevertheless, there seems to be a lack of a coherent framework facilitating the formulation of hypotheses, the testing of which could help resolve these discussions.

Executive Functions (EF), a set of domain-general cognitive processes that enable flexible, goal-directed behavior and regulate internal states in response to environmental demands (Diamond, 2013; Cristofori et al., 2019), arise as a promising candidate for application to insects’ behavioral neuroscience. EF are theorized to occupy a high-level regulatory role within the cognitive system, orchestrating lower-level processes to enable adaptive and context-sensitive behavior across changing environmental demands. Although the concept of EF emerged within human neuropsychology and is strongly associated with the function of the prefrontal cortex (PFC), recent comparative studies indicate that analogous, EF-like capabilities arise in vertebrates such as birds independently of cortical structures (Emery and Clayton, 2004). This suggests the potential of EF as a framework for describing cognitive processes on a more abstract level and as an organizing principle for understanding adaptive behavior across contexts, while nonetheless prompting questions about convergence and/or functional homology of anatomically differing structures.

This paper aims to position the EF framework as a promising perspective on insect behavior and cognition, entailing a distinctive set of characteristics that could be identified and tested, thus facilitating the development of conceptual and methodological tools for better exploring insects’ cognitive complexity.

First, we outline the EF framework’s origins in human psychology and neuroscience, followed by its adaptation to non-human animals. Subsequently, we move to insect cognition by reviewing cases where insect behavior has already been explicitly described in EF-related terms, and, most importantly, applying the framework de novo to reinterpret findings from studies that did not originally invoke EF-related concepts. While we emphasize the functional over structural characteristics of EF, we also consider potential neural substrates in insects that may support EF-like processes. Finally, we discuss the strengths and limitations of this perspective.

Executive functions are usually conceptualized as a set of processes at the top of the cognitive hierarchy (“metacognitive”) and are responsible for behavioral control, planning, problem-solving, achieving goals, and adapting to situational demands (Diamond, 2013; Cristofori et al., 2019). They are closely related to working memory (WM), attention, and cognitive flexibility. They are domain-general processes that influence all forms of cognition when control is required. EF impairments lead to deficits in non-automatic cognitive processes and behaviors. In humans, healthy development and effective performance of EF (Thompson and Steinbeis, 2020) are robust predictors of life outcomes, crucial for adaptation and decision-making (Moffitt et al., 2011). EF typically emerge in early childhood, strengthen through adolescence, peak in early adulthood, and decline with age, though not uniformly across all components (Lacreuse et al., 2020; Ferguson et al., 2021).

The EF concept is strongly linked to the clinical context, particularly to the neuropsychological perspectives on PFC dysfunction (Friedman and Robbins, 2022); however, the cybernetic tradition has played an important role in its development (Miller et al., 1968; Baddeley, 2010). The term EF is, to some extent, used interchangeably with cognitive control. Despite a well-established grounding in the PFC, EF are also associated with other anterior and posterior cortical regions (Bettcher et al., 2016).

Despite several categorizations of EF, one of the most widely used is a three-component model, comprising inhibition, shifting, and updating by Miyake et al. (2000). Inhibition refers to the process of suppressing an automatic reaction to a stimulus and/or suppressing task-irrelevant perceptions. Shifting refers to the process of exerting controlled changes between different tasks. Finally, updating refers to the process of keeping task-relevant information in one’s WM. Notably, EF categories sometimes overlap, reflecting varying research approaches and diverse terminology. In addition to just mentioned, the following concepts can be listed as examples of EF: cognitive flexibility, behavioral control, planning, decision-making, and problem-solving (Miyake et al., 2000; Cristofori et al., 2019).

Psychological concepts are often applied to research on other animals (Smith et al., 2003). Typically, such an application enables researchers to test the evolutionary roots of cognitive mechanisms and examine how shared or divergent neural architectures support similar functions (de Waal and Ferrari, 2010). This is exactly what happened in the case of research on EF in non-human primates, which focused on homologies with human PFC functions (Mansouri et al., 2017; Lacreuse et al., 2020). Over time, research expanded to include other mammals, notably domestic dogs, whose co-evolution with humans makes them valuable models for studying cognitive control in ecologically relevant contexts (Foraita et al., 2022). Parallel to these developments, rodents became key, though indirect, models in EF research due to their translational relevance for modeling human cognitive disorders (Bizon et al., 2012).

As exemplified by avian cognition research, EF concepts can be fruitfully applied to non-mammalian species. Corvids and psittacines–birds known for their broad behavioral repertoires and complex social interactions–have consistently shown proficiency in tasks probing EFs (Bobrowicz and Greiff, 2022). Crow species and African gray parrots, for instance, demonstrate advanced inhibitory control in detour-reaching tasks and delayed-rewards exchanges, as well as robust WM in object permanence and transposition paradigms (Hoffmann et al., 2011; Rössler and Auersperg, 2023). Reversal learning and multi-access puzzle tasks, often interpreted as indicators of cognitive flexibility, underscore the control capabilities of these taxa (Auersperg et al., 2011). Such findings challenge the centrality of cortical substrates to EF, as birds achieve functionally comparable outcomes, relying on pallial regions, including the nidopallium caudolaterale, which lacks the laminar organization of the mammalian neocortex yet seemingly supports comparable integrative processing (Emery and Clayton, 2004; Rose and Colombo, 2005; Kersten et al., 2024). This neuroanatomical divergence, coupled with functional convergence or homology, strengthens the case for defining EF not by structural correlates but by computational characteristics.

These findings have prompted inquiries into EF-like capacities across vertebrates, including species not typically characterized by large brains. Fish species such as guppies, cleaner wrasses, and zebrafish (Parker et al., 2013) have shown evidence of engaging in behaviors suggestive of EF. Guppies, for example, have succeeded in detour tasks requiring inhibition and spatial flexibility (Lucon-Xiccato et al., 2017; Gatto et al., 2018), while cleaner wrasses have demonstrated the ability to modulate their behavior based on the perceived perspective of their social partners (McAuliffe et al., 2021), indicating context-sensitive decision-making. Among recent findings are demonstrations of EF-like behavior in invertebrates, namely, cephalopod mollusks, particularly octopuses and cuttlefish. Despite their divergent neuroanatomy, cephalopods have shown advanced control and learning flexibility (Jozet-Alves et al., 2013). Cuttlefish, for example, have been shown to forgo an immediate but less-preferred rewards in favor of a delayed, higher-value option–a clear functional parallel to human delay-of-gratification paradigms (Schnell et al., 2021b). These behaviors have been further correlated with performance in reversal learning paradigms, suggesting an integrated system of inhibitory control and flexibility. Similarly, octopuses have demonstrated the ability to adapt to changing problem-solving contexts and to remember prior outcomes in decision tasks (Schnell et al., 2021a). These findings are especially significant in the discussed context, given the absence of any directly homologous structures to the vertebrate cortex.

Against the background just described, we propose taking the next step–applying the EF concepts to insects, whose neural architecture is extremely miniaturized. Our proposal is not made in a vacuum and may not appear entirely original, as several insect studies have already used terms associated with EF. Thus, in Section “4.1 Verbatim usage,” we briefly review the verbatim use of EF-related terminology, focusing primarily on studies involving bees and ants. While these studies provide a credible starting point, some of them do not situate individual terms within the broader framework of EF. This is evident in the lack of cross-referencing between specific terms (e.g., the absence of any indication that inhibition and updating may reflect distinct yet related capacities), the omission of references to foundational authors or key publications on EF, and–notably–the absence of the term “executive functions” itself. These limitations are important, as the terms used in the existing literature do not necessarily denote EF as typically understood in neuropsychology. Before proposing means to clarify whether a particular capacity in insects sufficiently resembles EF in vertebrates to warrant such a label, in Section “4.2 Functional inference” we draw attention to a somewhat reversed situation–namely, cases in which certain processes are not explicitly labeled as EF, even when their executive nature could be readily inferred. Next, in Section “4.3 Conceptual integration,” we describe another, more interpretively challenging context in which EF may be involved in the execution of cognitive tasks, for example, as is the case of tasks designed to assess numerosity processing, where an animal must inhibit conflicting perceptual cues in favor of numerical information.

To avoid turning EF into a catch-all label for complex cognition in insects, it is important to identify cognitive phenomena that, while complex, are not executive functions themselves. Following Knauff and Wolf (2010), by complex cognition we mean that which “takes place under complex conditions in which a multitude of cognitive processes interact with one another or with other non-cognitive processes” (p. 100). Therefore, phenomena such as reasoning, or knowledge use and transfer are examples of complex cognition, but they are not EFs themselves (even if there is a possibility of interaction between them). Here, we briefly present a handful of examples of complex cognition in insects that should not be conflated with EF.

Ants have been observed to learn new object affordances and to act according to acquired knowledge (Poissonnier et al., 2023). This exemplifies a complex cognition, constituted by interactions among perception, memory, and social learning, that might interact with EF (e.g., inhibition might be needed to stop an initial reaction), but it is not EF itself. Next, Ebina and Mizunami (2020) found that crickets prefer food sources where a living conspecific had been encountered and avoid those associated with dead conspecifics. This presents a form of reasoning: “If a conspecific is alive, then the food is safe; if it is dead, the food is dangerous.” However, EF processes are not required for this behavior to occur. Finally, in a study by Solvi et al. (2020), bumblebees performed well in a task requiring cross-modal (in this case, visual and tactile) object recognition. The insects were able to recognize objects presented in a different sensory modality from the one used during the learning phase. This ability represents a form of object abstraction, but does not imply the involvement of EF.

Although EFs do not necessarily depend on the presence of a cerebral cortex, the successful application of the EF framework requires identifying the underlying neural substrate.

Two insect brain structures consistently emerge in the context of complex cognition: the mushroom bodies (MBs) and the central complex (CX) (Heisenberg, 2003; Pfeiffer and Homberg, 2014; Barron and Klein, 2016). MBs are bilaterally symmetrical structures crucial for associative learning, memory formation, and sensory integration (Heisenberg, 2003; Farris, 2011). Lesion and imaging studies have shown that MBs are indispensable for tasks requiring the formation of stimulus–rewards contingencies and behavioral flexibility (Zars, 2000; Menzel, 2012). Importantly, the MBs receive multimodal sensory input and contribute to state-dependent behavioral modulation–characteristics reminiscent of the integrative role played by the PFC in vertebrates (Strausfeld et al., 1998; Barron and Klein, 2016).

Central complex, on the other hand, is a singular structure located in the midline of the insect brain, involved in spatial orientation, motor control, and decision-making (Strausfeld and Hirth, 2013; Pfeiffer and Homberg, 2014; Honkanen et al., 2019). It appears particularly important for action selection and goal-directed navigation, functioning through the integration of internal states and external cues to orchestrate context-appropriate responses (Turner-Evans and Jayaraman, 2016; Honkanen et al., 2019; Goulard et al., 2023). In the fruit fly, CX integrates internal variables such as sleep homeostasis (Flores-Valle et al., 2021) and path-integrated heading direction to modulate navigational decisions, demonstrating internal control over spatially guided actions (Seelig and Jayaraman, 2015; Kottler et al., 2019).

These findings suggest that the MBs and CX form part of a distributed control architecture in which executive-like functions may emerge (Menzel, 2012; Pfeiffer and Homberg, 2014; Collett and Collett, 2018).

We propose that several insect behaviors can be systematically interpreted through the EF framework–an approach commonly used in vertebrate cognitive and behavioral neuroscience research. The broader application of the EF framework can help advance insect cognition research in numerous ways. First, it provides a standardized vocabulary for identifying and categorizing behavior dependent on inhibition, WM, and cognitive flexibility–capacities already investigated across a wide range of animals, thus facilitating cross-species comparisons. Next, since many of these concepts have already been used in the insect cognition literature–albeit often without a broader theoretical context–adopting the EF framework would allow for both retrospective reinterpretation of existing findings and unification of terminology in future research in an otherwise fragmented field.

Furthermore, adopting the EF framework opens new research avenues by highlighting developmental patterns and constraints that are characteristic of executive control across taxa (Lacreuse et al., 2020). If insects indeed exhibit EF-like capacities, comparable trajectories, though likely compressed and mechanistically distinct from the vertebrate ones, may be observable. As previously proposed, if the MBs and CX serve as primary substrates for executive-like functions, then it would be expected that developmental changes in their structure could correlate with corresponding shifts in EF-related behavior (Strausfeld and Hirth, 2013; Collett and Collett, 2018). When examining potential trajectories, a sharp distinction must be drawn between hemimetabolous and holometabolous species, due to fundamental differences in their neural ontogeny. In hemimetabolous insects, MBs are known to exhibit incremental development across successive juvenile instars (Farris, 2011; Strausfeld and Hirth, 2013), raising the possibility of a more vertebrate-like pattern of gradual maturation in executive capacities. However, these species remain behaviorally understudied in this context, and systematic links between MB development and EF-like behavior are currently lacking. In contrast, holometabolous insects undergo extensive neural reorganization during metamorphosis, rapidly achieving mature-like neural structures post-eclosion. In these species, subsequent plasticity arises primarily from environmental modulation, most notably, experience-dependent expansion of the MB neuropil, as observed in honeybees transitioning from in-nest tasks to foraging (Menzel, 2012; Collett and Collett, 2018). While this indicates functional refinement post-emergence, it does not imply a prolonged developmental arc akin to the vertebrate prefrontal cortex. Nevertheless, some insect species do exhibit patterns suggestive of developmental and age-related modulation of EF-like behavior, such as cognitive decline in older individuals (Münch et al., 2013), which bears significant resemblance to vertebrate EF dynamics.

Moreover, EF tasks often reveal inherent trade-offs between speed and accuracy, context-dependent modulation of behavior, and limitations in attentional or memory resources–features that distinguish executive processing from more rigid forms of associative learning (Ye et al., 2019; Ibbotson, 2023). Recognizing these hallmarks provides a basis for designing behavioral paradigms that probe not only whether insects can perform a task, but also how their performance varies with internal states, task demands, and conflicting goals. Such refinement may enable deeper cross-species comparisons, including parallel task designs for insects and vertebrates.

However, applying the EF framework to insects entails significant challenges. Notably, numerous EF paradigms used in humans rely on verbal instruction, explicit goal comprehension, and introspective reporting–none of which are accessible in studies involving insect models. EF-related terminology can be easily overextended. Caution is especially important, given that superficially similar outcomes can be driven by distinct cognitive processes (Gatto et al., 2023). Thus, interpreting insect behavior in terms of EF is challenging, particularly when attempting to distinguish between genuinely flexible processes and highly specialized, context-dependent associative mechanisms (Strelevitz et al., 2024). Behaviors that appear adaptive or strategic may, in fact, emerge from domain-specific heuristics rather than from executive functioning.

The EF framework should not become a catch-all label for complex cognition. Its strength lies in its ability to differentiate latent control processes based on characteristic signatures, not merely on specific behavioral outcomes. Despite its limitations, the EF framework offers a powerful comparative tool, encouraging integrative research across behavior, neurobiology, and modeling.