Hierarchical active inference for cognitive architectures

This thesis explores the methods based on the free energy principle and active inference for modelling cognition. Active inference is an emerging framework for designing intelligent agents where psychological processes are cast in terms of Bayesian inference. Here, I appeal to it to test the design of a set of cognitive architectures, via simulation. These architectures are defined in terms of generative models where an agent executes a task under the assumption that all cognitive processes aspire to the same objective: the minimization of variational free energy. Chapter 1 introduces the free energy principle and its assumptions about self-organizing systems. Chapter 2 describes how from the mechanics of self-organization can emerge a minimal form of cognition able to achieve autopoiesis. In chapter 3 I present the method of how I formalize generative models for action and perception. The architectures proposed allow providing a more biologically plausible account of more complex cognitive processing that entails deep temporal features. I then present three simulation studies that aim to show different aspects of cognition, their associated behavior and the underlying neural dynamics. In chapter 4, the first study proposes an architecture that represents the visuomotor system for the encoding of actions during action observation, understanding and imitation. In chapter 5, the generative model is extended and is lesioned to simulate brain damage and neuropsychological patterns observed in apraxic patients. In chapter 6, the third study proposes an architecture for cognitive control and the modulation of attention for action selection. At last, I argue how active inference can provide a formal account of information processing in the brain and how the adaptive capabilities of the simulated agents are a mere consequence of the architecture of the generative models. Cognitive processing, then, becomes an emergent property of the minimization of variational free energy.

Interactions between normative systems and software cognitive agents. A formalization in temporal modal defeasible logic and its implementation

Sustainable computer systems require some flexibility to adapt to environmental unpredictable changes. A solution lies in autonomous software agents which can adapt autonomously to their environments. Though autonomy allows agents to decide which behavior to adopt, a disadvantage is a lack of control, and as a side effect even untrustworthiness: we want to keep some control over such autonomous agents. How to control autonomous agents while respecting their autonomy? A solution is to regulate agents’ behavior by norms. The normative paradigm makes it possible to control autonomous agents while respecting their autonomy, limiting untrustworthiness and augmenting system compliance. It can also facilitate the design of the system, for example, by regulating the coordination among agents. However, an autonomous agent will follow norms or violate them in some conditions. What are the conditions in which a norm is binding upon an agent? While autonomy is regarded as the driving force behind the normative paradigm, cognitive agents provide a basis for modeling the bindingness of norms. In order to cope with the complexity of the modeling of cognitive agents and normative bindingness, we adopt an intentional stance. Since agents are embedded into a dynamic environment, things may not pass at the same instant. Accordingly, our cognitive model is extended to account for some temporal aspects. Special attention is given to the temporal peculiarities of the legal domain such as, among others, the time in force and the time in efficacy of provisions. Some types of normative modifications are also discussed in the framework. It is noteworthy that our temporal account of legal reasoning is integrated to our commonsense temporal account of cognition. As our intention is to build sustainable reasoning systems running unpredictable environment, we adopt a declarative representation of knowledge. A declarative representation of norms will make it easier to update their system representation, thus facilitating system maintenance; and to improve system transparency, thus easing system governance. Since agents are bounded and are embedded into unpredictable environments, and since conflicts may appear amongst mental states and norms, agent reasoning has to be defeasible, i.e. new pieces of information can invalidate formerly derivable conclusions. In this dissertation, our model is formalized into a non-monotonic logic, namely into a temporal modal defeasible logic, in order to account for the interactions between normative systems and software cognitive agents.

Mathematical models of cognitive processes

The research activity carried out during the PhD course was focused on the development of mathematical models of some cognitive processes and their validation by means of data present in literature, with a double aim: i) to achieve a better interpretation and explanation of the great amount of data obtained on these processes from different methodologies (electrophysiological recordings on animals, neuropsychological, psychophysical and neuroimaging studies in humans), ii) to exploit model predictions and results to guide future research and experiments. In particular, the research activity has been focused on two different projects: 1) the first one concerns the development of neural oscillators networks, in order to investigate the mechanisms of synchronization of the neural oscillatory activity during cognitive processes, such as object recognition, memory, language, attention; 2) the second one concerns the mathematical modelling of multisensory integration processes (e.g. visual-acoustic), which occur in several cortical and subcortical regions (in particular in a subcortical structure named Superior Colliculus (SC)), and which are fundamental for orienting motor and attentive responses to external world stimuli. This activity has been realized in collaboration with the Center for Studies and Researches in Cognitive Neuroscience of the University of Bologna (in Cesena) and the Department of Neurobiology and Anatomy of the Wake Forest University School of Medicine (NC, USA). PART 1. Objects representation in a number of cognitive functions, like perception and recognition, foresees distribute processes in different cortical areas. One of the main neurophysiological question concerns how the correlation between these disparate areas is realized, in order to succeed in grouping together the characteristics of the same object (binding problem) and in maintaining segregated the properties belonging to different objects simultaneously present (segmentation problem). Different theories have been proposed to address these questions (Barlow, 1972). One of the most influential theory is the so called “assembly coding”, postulated by Singer (2003), according to which 1) an object is well described by a few fundamental properties, processing in different and distributed cortical areas; 2) the recognition of the object would be realized by means of the simultaneously activation of the cortical areas representing its different features; 3) groups of properties belonging to different objects would be kept separated in the time domain. In Chapter 1.1 and in Chapter 1.2 we present two neural network models for object recognition, based on the “assembly coding” hypothesis. These models are networks of Wilson-Cowan oscillators which exploit: i) two high-level “Gestalt Rules” (the similarity and previous knowledge rules), to realize the functional link between elements of different cortical areas representing properties of the same object (binding problem); 2) the synchronization of the neural oscillatory activity in the γ-band (30-100Hz), to segregate in time the representations of different objects simultaneously present (segmentation problem). These models are able to recognize and reconstruct multiple simultaneous external objects, even in difficult case (some wrong or lacking features, shared features, superimposed noise). In Chapter 1.3 the previous models are extended to realize a semantic memory, in which sensory-motor representations of objects are linked with words. To this aim, the network, previously developed, devoted to the representation of objects as a collection of sensory-motor features, is reciprocally linked with a second network devoted to the representation of words (lexical network) Synapses linking the two networks are trained via a time-dependent Hebbian rule, during a training period in which individual objects are presented together with the corresponding words. Simulation results demonstrate that, during the retrieval phase, the network can deal with the simultaneous presence of objects (from sensory-motor inputs) and words (from linguistic inputs), can correctly associate objects with words and segment objects even in the presence of incomplete information. Moreover, the network can realize some semantic links among words representing objects with some shared features. These results support the idea that semantic memory can be described as an integrated process, whose content is retrieved by the co-activation of different multimodal regions. In perspective, extended versions of this model may be used to test conceptual theories, and to provide a quantitative assessment of existing data (for instance concerning patients with neural deficits). PART 2. The ability of the brain to integrate information from different sensory channels is fundamental to perception of the external world (Stein et al, 1993). It is well documented that a number of extraprimary areas have neurons capable of such a task; one of the best known of these is the superior colliculus (SC). This midbrain structure receives auditory, visual and somatosensory inputs from different subcortical and cortical areas, and is involved in the control of orientation to external events (Wallace et al, 1993). SC neurons respond to each of these sensory inputs separately, but is also capable of integrating them (Stein et al, 1993) so that the response to the combined multisensory stimuli is greater than that to the individual component stimuli (enhancement). This enhancement is proportionately greater if the modality-specific paired stimuli are weaker (the principle of inverse effectiveness). Several studies have shown that the capability of SC neurons to engage in multisensory integration requires inputs from cortex; primarily the anterior ectosylvian sulcus (AES), but also the rostral lateral suprasylvian sulcus (rLS). If these cortical inputs are deactivated the response of SC neurons to cross-modal stimulation is no different from that evoked by the most effective of its individual component stimuli (Jiang et al 2001). This phenomenon can be better understood through mathematical models. The use of mathematical models and neural networks can place the mass of data that has been accumulated about this phenomenon and its underlying circuitry into a coherent theoretical structure. In Chapter 2.1 a simple neural network model of this structure is presented; this model is able to reproduce a large number of SC behaviours like multisensory enhancement, multisensory and unisensory depression, inverse effectiveness. In Chapter 2.2 this model was improved by incorporating more neurophysiological knowledge about the neural circuitry underlying SC multisensory integration, in order to suggest possible physiological mechanisms through which it is effected. This endeavour was realized in collaboration with Professor B.E. Stein and Doctor B. Rowland during the 6 months-period spent at the Department of Neurobiology and Anatomy of the Wake Forest University School of Medicine (NC, USA), within the Marco Polo Project. The model includes four distinct unisensory areas that are devoted to a topological representation of external stimuli. Two of them represent subregions of the AES (i.e., FAES, an auditory area, and AEV, a visual area) and send descending inputs to the ipsilateral SC; the other two represent subcortical areas (one auditory and one visual) projecting ascending inputs to the same SC. Different competitive mechanisms, realized by means of population of interneurons, are used in the model to reproduce the different behaviour of SC neurons in conditions of cortical activation and deactivation. The model, with a single set of parameters, is able to mimic the behaviour of SC multisensory neurons in response to very different stimulus conditions (multisensory enhancement, inverse effectiveness, within- and cross-modal suppression of spatially disparate stimuli), with cortex functional and cortex deactivated, and with a particular type of membrane receptors (NMDA receptors) active or inhibited. All these results agree with the data reported in Jiang et al. (2001) and in Binns and Salt (1996). The model suggests that non-linearities in neural responses and synaptic (excitatory and inhibitory) connections can explain the fundamental aspects of multisensory integration, and provides a biologically plausible hypothesis about the underlying circuitry.

Motor control system in Parkinson’s disease: a modeling approach

Parkinson’s disease is a neurodegenerative disorder due to the death of the dopaminergic neurons of the substantia nigra of the basal ganglia. The process that leads to these neural alterations is still unknown. Parkinson’s disease affects most of all the motor sphere, with a wide array of impairment such as bradykinesia, akinesia, tremor, postural instability and singular phenomena such as freezing of gait. Moreover, in the last few years the fact that the degeneration in the basal ganglia circuitry induces not only motor but also cognitive alterations, not necessarily implicating dementia, and that dopamine loss induces also further implications due to dopamine-driven synaptic plasticity got more attention. At the present moment, no neuroprotective treatment is available, and even if dopamine-replacement therapies as well as electrical deep brain stimulation are able to improve the life conditions of the patients, they often present side effects on the long term, and cannot recover the neural loss, which instead continues to advance. In the present thesis both motor and cognitive aspects of Parkinson’s disease and basal ganglia circuitry were investigated, at first focusing on Parkinson’s disease sensory and balance issues by means of a new instrumented method based on inertial sensor to provide further information about postural control and postural strategies used to attain balance, then applying this newly developed approach to assess balance control in mild and severe patients, both ON and OFF levodopa replacement. Given the inability of levodopa to recover balance issues and the new physiological findings than underline the importance in Parkinson’s disease of non-dopaminergic neurotransmitters, it was therefore developed an original computational model focusing on acetylcholine, the most promising neurotransmitter according to physiology, and its role in synaptic plasticity. The rationale of this thesis is that a multidisciplinary approach could gain insight into Parkinson’s disease features still unresolved.

Bridging the gap:exploring the cognitive impact of InterpretBank on Chinese interpreting trainees

This exploratory research project developed a cognitive situated approach to studying aspects of simultaneous interpreting with quantitative, confirmatory methods. To do so, it explored how to determine the potential benefits of using a computer-assisted interpreting tool, InterpretBank, among 22 Chinese interpreting trainees with Chinese L1 and English L2. The informants were mostly 2nd-year female students with an average age of 24.7 enrolled in Chinese MA interpreting programs. The study adopted a pretest and posttest design with three cycles. The independent variable was using Excel or InterpretBank. After Cycle I (pre-test), the sample split into control (Excel) and experimental (InterpretBank) groups. Tool choice was compulsory in Cycle II but not Cycle III. The source materials for each cycle were pairs of matching transcripts from popular science podcasts. Informants compiled glossaries out of one transcript, while the other one was edited for simultaneous interpreting, with 39 terms as potential problem triggers. Quantitative profiling results showed that InterpretBank informants spent less time on glossary compilation, generated more terms faster than Excel informants, but their glossaries were less diverse (personal) and longer. The booth tasks yielded no significant differences in fluency indicators except for more bumps (200-600ms silent time gaps) for InterpretBank in Cycle II. InterpretBank informants had more correct renditions in Cycles II and III but there was no statistically significant difference among accuracy indicators per cycle. Holistic quality assessments by PhD raters showed InterpretBank consistently outperforming Excel, suggesting a positive InterpretBank impact on SI quality. However, some InterpretBank implementations raised cognitive ergonomic concerns for Chinese, potentially undermining its utility. Overall, results were mixed regarding InterpretBank benefits for Chinese trainees, but the project was successful in developing cognitive situated interpreting study methods, constructs and indicators.