Self-supervised learning has become an increasingly important paradigm in the domain of machine intelligence. Furthermore, evidence for self-supervised adaptation, such as contrastive formulations, has emerged in recent computational neuroscience and brain-inspired research. Nevertheless, current work on self-supervised learning relies on biologically implausible credit assignment -- in the form of backpropagation of errors -- and feedforward inference, typically a forward-locked pass. Predictive coding, in its mechanistic form, offers a biologically plausible means to sidestep these backprop-specific limitations. However, unsupervised predictive coding rests on learning a generative model of raw pixel input (akin to ``generative AI'' approaches), which entails predicting a potentially high dimensional input; on the other hand, supervised predictive coding, which learns a mapping between inputs to target labels, requires human annotation, and thus incurs the drawbacks of supervised learning. In this work, we present a scheme for self-supervised learning within a neurobiologically plausible framework that appeals to the free energy principle, constructing a new form of predictive coding that we call meta-representational predictive coding (MPC). MPC sidesteps the need for learning a generative model of sensory input (e.g., pixel-level features) by learning to predict representations of sensory input across parallel streams, resulting in an encoder-only learning and inference scheme. This formulation rests on active inference (in the form of sensory glimpsing) to drive the learning of representations, i.e., the representational dynamics are driven by sequences of decisions made by the model to sample informative portions of its sensorium.
Synopsis Apomorphine, a non-selective dopamine agonist, has been used as a pharmacological probe for investigating central dopaminergic neurotransmission in psychiatric illness. In this study repeated measurements of regional cerebral blood flow (rCBF) were made in normal volunteers before, and after, the administration of apomorphine (5 or 10 μg/kg), or placebo. The difference in rCBF, before and after drug (apomorphine versus placebo), was used to identify brain areas affected by apomorphine. Compared to placebo, both doses of apomorphine increased blood flow in the anterior cingulate cortex. Apomorphine 10 μg/kg also increased prefrontal rCBF (right > left). No decreases in rCBF were noted following either dose of apomorphine. Apomorphine-induced increases of anterior cingulate blood flow might serve as an in vivo index of central dopamine function. Such an approach would complement established neuroendocrine challenge paradigms for investigating central dopamine neurotransmission in psychiatric illness.
This paper presents a nonlinear principal component analysis (PCA) that identifies underlying sources causing the expression of spatial modes or patterns of activity in neuroimaging time–series. The critical aspect of this technique is that, in relation to conventional PCA, the sources can interact to produce (second–order) spatial modes that represent the modulation of one (first–order) spatial mode by another. This nonlinear PCA uses a simple neural network architecture that embodies a specific form for the nonlinear mixing of sources that cause observed data. This form is motivated by a second–order approximation to any general nonlinear mixing and emphasizes interactions among pairs of sources. By introducing these nonlinearities principal components obtain with a unique rotation and scaling that does not depend on the biologically implausible constraints adopted by conventional PCA. The technique is illustrated by application to functional (positron emission tomography and functional magnetic resonance imaging) imaging data where the ensuing first– and second–order modes can be interpreted in terms of distributed brain systems. The interactions among sources render the expression of any one mode context–sensitive, where that context is established by the expression of other modes. The examples considered include interactions between cognitive states and time (i.e. adaptation or plasticity in PETdata) and among functionally specialized brain systems (using a fMRI study of colour and motion processing).
We present a method for detecting event-related responses in functional magnetic resonance imaging (fMRI). The occurrence of time-locked activations is formulated in terms of the general linear model, i.e., multiple linear regression. This permits the use of established statistical techniques that correct for multiple comparisons in the context of spatially smooth and serially correlated data. Responses are modelled using event-related temporal basis functions. Inferences are then made about all components of the model, using the F-ratio at all voxels in the image, to produce a statistical parametric map (SPM{F}). This method allows for the experimental design to relate the timing of events to the acquisition of data to give a temporal resolution (with respect to the event-related response) far better than the scanning repeat time. Hum. Brain Mapping 5:243–248, 1997. © 1997 Wiley-Liss, Inc.
This paper offers a prospectus of what might be achievable in the development of emotional recognition devices. It provides a conceptual overview of the free energy principle; including Markov blankets, active inference, and – in particular – a discussion of selfhood and theory of mind, followed by a brief explanation of how these concepts can explain both neural and cultural models of emotional inference. The underlying hypothesis is that emotion recognition and inference devices will evolve from state-of-the-art deep learning models into active inference schemes that go beyond marketing applications and become adjunct to psychiatric practice. Specifically, this paper proposes that a second wave of emotion recognition devices will be equipped with an emotional lexicon (or the ability to epistemically search for one), allowing the device to resolve uncertainty about emotional states by actively eliciting responses from the user and learning from these responses. Following this, a third wave of emotional devices will converge upon the user’s generative model, resulting in the machine and human engaging in a reciprocal, prosocial emotional interaction, i.e. sharing a generative model of emotional states.
