Unlimited Associative Learning and the origins of consciousness: a primer and some predictions

Birch, Jonathan; Ginsburg, Simona; Jablonka, Eva

doi:10.1007/s10539-020-09772-0

Cited by 98 publications

(68 citation statements)

References 65 publications

(76 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The current account may also provide a fuller picture of why ‘unlimited associative learning’—due to facilitating the complex reafferent processes associated with a deep self-model—may be a marker of the evolutionary transition to minimal consciousness ( Bronfman et al. 2016 ; Ginsburg and Jablonka 2019 ; Birch et al. 2020 ).…”

Section: Consciousness In Other Systemsmentioning

confidence: 95%

Consciousness in active inference: Deep self-models, other minds, and the challenge of psychedelic-induced ego-dissolution

Deane

2021

Neuroscience of Consciousness

View full text Add to dashboard Cite

Predictive processing approaches to brain function are increasingly delivering promise for illuminating the computational underpinnings of a wide range of phenomenological states. It remains unclear, however, whether predictive processing is equipped to accommodate a theory of consciousness itself. Furthermore, objectors have argued that without specification of the core computational mechanisms of consciousness, predictive processing is unable to inform the attribution of consciousness to other non-human (biological and artificial) systems. In this paper, I argue that an account of consciousness in the predictive brain is within reach via recent accounts of phenomenal self-modelling in the active inference framework. The central claim here is that phenomenal consciousness is underpinned by ‘subjective valuation’—a deep inference about the precision or ‘predictability’ of the self-evidencing (‘fitness-promoting’) outcomes of action. Based on this account, I argue that this approach can critically inform the distribution of experience in other systems, paying particular attention to the complex sensory attenuation mechanisms associated with deep self-models. I then consider an objection to the account: several recent papers argue that theories of consciousness that invoke self-consciousness as constitutive or necessary for consciousness are undermined by states (or traits) of ‘selflessness’; in particular the ‘totally selfless’ states of ego-dissolution occasioned by psychedelic drugs. Drawing on existing work that accounts for psychedelic-induced ego-dissolution in the active inference framework, I argue that these states do not threaten to undermine an active inference theory of consciousness. Instead, these accounts corroborate the view that subjective valuation is the constitutive facet of experience, and they highlight the potential of psychedelic research to inform consciousness science, computational psychiatry and computational phenomenology.

show abstract

Section: Consciousness In Other Systemsmentioning

confidence: 95%

Consciousness in active inference: Deep self-models, other minds, and the challenge of psychedelic-induced ego-dissolution

Deane

2021

Neuroscience of Consciousness

View full text Add to dashboard Cite

show abstract

“…Further, principles of association may be surprisingly powerful if they are capable of representing specific relational structures as particular graphs/networks, which are increasingly being recognized as powerful learning and inferential systems (Gentner, 2010;Zhou et al, 2019;Crouse et al, 2020). Some have even suggested that the spatial and temporal mapping abilities of the H/E-S represent a special case of general purpose unlimited associative learning (Birch et al, 2020), such as aggregation across episodes (Mack et al, 2018(Mack et al, , 2020Mok and Love, 2019), or of dynamically evolving graphs capable of generating cognitive maps via representational "cloning" (Gothoskar et al, 2019). The central role of the H/E-S for higher-order cognition is further understandable in light of the fact that many (and possibly most) aspects of intelligence can be described as search processes (Conant and Ashby, 1970;Hills et al, 2010), which might be even more clearly apparent if we think of the possibility of spatializing abstract domains such as complex feature spaces (Eichenbaum, 2015;Whittington et al, 2018), or even time (Howard, 2018;Gauthier et al, 2019).…”

Section: H/e-s As Orchestrator Of High-level Cognitionmentioning

confidence: 99%

Generalized Simultaneous Localization and Mapping (G-SLAM) as unification framework for natural and artificial intelligences: towards reverse engineering the hippocampal/entorhinal system and principles of high-level cognition

Safron¹,

Çatal²,

Verbelen³

2021

Preprint

View full text Add to dashboard Cite

Simultaneous localization and mapping (SLAM) represents a fundamental problem for autonomous embodied systems, for which hippocampal/entorhinal system (H/E-S) adaptations have been optimized over the course of evolution. We have developed a biologically-inspired SLAM architecture based on latent variable generative modeling within the Free Energy Principle and Active Inference (FEP-AI) framework, which affords flexible navigation and planning in mobile robots. We have primarily focused on attempting to reverse engineer H/E-S ‘design’ properties, but here we consider ways in which SLAM principles from robotics may help us better understand nervous systems and emergent minds. After reviewing LatentSLAM and notable features of this control architecture, we consider how the H/E-S may realize these functional properties not only for physical navigation, but also with respect to high-level cognition understood as generalized simultaneous localization and mapping (G-SLAM). We focus on loop closure, graph relaxation, and node duplication as particularly impactful architectural features, suggesting these computational phenomena may contribute to understanding cognitive insight (as proto-causal-inference), accommodation (as integration into existing schemas), and assimilation (as category formation). All these operations can similarly be describable in terms of structure/category learning on multiple levels of abstraction. However, here we adopt an ecological rationality perspective, framing H/E-S functions as orchestrating SLAM processes within both concrete and abstract hypothesis spaces. In this navigation/search process, adaptive cognitive equilibration between assimilation and accommodation involves balancing tradeoffs between exploration and exploitation; this dynamic equilibrium may be near optimally achieved in FEP-AI agents, wherein control systems governed by expected free energy objective functions naturally balance model simplicity and accuracy. With respect to structure learning, such a balance would involve constructing models and categories that are neither too inclusive nor exclusive. We propose these (generalized) SLAM phenomena may represent some of the most impactful sources of variation in cognition both within and between individuals, suggesting the impacts of these neuromodulators on H/E-S functioning may potentially illuminate the adaptive significance of these signaling pathways as fundamental cybernetic control parameters. Finally, we discuss how understanding H/E-S contributions to G-SLAM may provide a unifying framework for high-level cognition and its potential realization in artificial intelligences.

show abstract

“…Our definition of “complex operant learning” was initially rather vague, but it came to mean learning many new survival behaviors from experience based on rewards and punishments [ 72 ]. Now, in my opinion, it is stated even better in the Unlimited Associative Learning (UAL) concept of Ginsburg and Jablonka [ 41 ], which was recently upgraded [ 90 ] as follows: UAL says an animal is conscious if it can learn from compound, novel stimuli; learn from what it already has learned and then do so again and again; learn by trace conditioning; and perform “valence switching,” meaning the animal can learn to like (approach) what it formerly disliked (avoided) and vice versa.…”

Section: Neurobiological Naturalism and Iitmentioning

confidence: 99%

A Traditional Scientific Perspective on the Integrated Information Theory of Consciousness

Mallatt

2021

Entropy

View full text Add to dashboard Cite

This paper assesses two different theories for explaining consciousness, a phenomenon that is widely considered amenable to scientific investigation despite its puzzling subjective aspects. I focus on Integrated Information Theory (IIT), which says that consciousness is integrated information (as ϕMax) and says even simple systems with interacting parts possess some consciousness. First, I evaluate IIT on its own merits. Second, I compare it to a more traditionally derived theory called Neurobiological Naturalism (NN), which says consciousness is an evolved, emergent feature of complex brains. Comparing these theories is informative because it reveals strengths and weaknesses of each, thereby suggesting better ways to study consciousness in the future. IIT’s strengths are the reasonable axioms at its core; its strong logic and mathematical formalism; its creative “experience-first” approach to studying consciousness; the way it avoids the mind-body (“hard”) problem; its consistency with evolutionary theory; and its many scientifically testable predictions. The potential weakness of IIT is that it contains stretches of logic-based reasoning that were not checked against hard evidence when the theory was being constructed, whereas scientific arguments require such supporting evidence to keep the reasoning on course. This is less of a concern for the other theory, NN, because it incorporated evidence much earlier in its construction process. NN is a less mature theory than IIT, less formalized and quantitative, and less well tested. However, it has identified its own neural correlates of consciousness (NCC) and offers a roadmap through which these NNCs may answer the questions of consciousness using the hypothesize-test-hypothesize-test steps of the scientific method.

show abstract

Unlimited Associative Learning and the origins of consciousness: a primer and some predictions

Cited by 98 publications

References 65 publications

Consciousness in active inference: Deep self-models, other minds, and the challenge of psychedelic-induced ego-dissolution

Consciousness in active inference: Deep self-models, other minds, and the challenge of psychedelic-induced ego-dissolution

Generalized Simultaneous Localization and Mapping (G-SLAM) as unification framework for natural and artificial intelligences: towards reverse engineering the hippocampal/entorhinal system and principles of high-level cognition

A Traditional Scientific Perspective on the Integrated Information Theory of Consciousness

Contact Info

Product

Resources

About