The attempt to provide a firm scientific basis for understanding consciousness is now in full swing, with special contributions from two areas. One is experimental: brain imaging is providing ever increasing detail of the brain structures used by humans (and other animals) as they solve a variety of tasks, including those of higher cognition. The other is theoretical: the discipline of neural networks is allowing models of these cognitive processes to be constructed and tested against the available data. In particular, a control framework can be created to give a global view of the brain. The highest cognitive process, that of consciousness, is naturally a target for such experimentation and modelling. This paper reviews available data and related models leading to the central representation, which involves particular brain regions and functional processing. Principles of consciousness, which have great relevance to the question in the title, are thereby deduced. The requisite neuronal systems needed to provide animal experience, and the problem of assessing the quality and quantity of such experience, will then be considered. In conclusion, animal consciousness is seen to exist broadly across those species with the requisite control structures; the level of pain and other sensations depends in an increasingly well-defined manner on the complexity of the cerebral apparatus.

GABA (gamma-amino-butyric acid) is the predominant neurotransmitter in the mammalian suprachiasmatic nucleus (SCN), with a central role in circadian time-keeping. We therefore undertook an ultrastructural analysis of the GABA-containing innervation in the SCN of mice and rats using immunoperoxidase and immunogold procedures. GABA-immunoreactive (GABA-ir) neurons were identified by use of anti-GABA and anti-GAD (glutamic acid decarboxylase) antisera. The relationship between GABA-ir elements and the most prominent peptidergic neurons in the SCN, containing vasopressin-neurophysin (VP-NP) or vasoactive intestinal polypeptide (VIP), was also studied. Within any given field in the SCN, approximately 40–70% of the neuronal profiles were GABA-ir. In GABA-ir somata, immunogold particles were prominent over mitochondria, sparse over cytoplasm, and scattered as aggregates over nucleoplasm. In axonal boutons, gold particles were concentrated over electron-lucent synaptic vesicles (diameter 40–60 nm) and mitochondria, and in some instances over dense-cored vesicles (DCVs, diameter 90–110 nm). GABA-ir boutons formed either symmetric or asymmetric synaptic contacts with somata, dendritic shafts and spines, and occasionally with other terminals (axo-axonic). Homologous or autaptic connections (GABA on GABA, or GAD on GAD) were common. Although GABA appeared to predominate in most neuronal profiles, colocalisation of GABA within neurons that were predominantly neuropeptide-containing was also evident. About 66% of the VIP-containing boutons and 32% of the vasopressinergic boutons contained GABA. The dense and complex GABAergic network that pervades the SCN is therefore comprised of multiple neuronal phenotypes containing GABA, including a wide variety of axonal boutons that impinge on heterologous and homologous postsynaptic sites.