'System' (from sunistanai, 'to combine' in Greek) is abundant in colloquial 'scientinglish'. 'Let me tell you about my system' or 'did you try it in your system' are only selected examples of ritualistic utterances of scientific "culture. An experimental system well-matched to the research goal is a key to successful research programmes. Prolific systems facilitate the road to academic tenure and fame, and trendy systems increase the probability of acceptance of manuscripts into respectable scientific journals. The popularity of 'system', therefore, deserves an attempt to put some order to the system.
What are systems? In the most general sense all systems are sets of interconnected units (definition 1). These units may be either concrete (e.g. parts of a machine) or abstract (e.g. concepts in a theory). In some cases the system is delineated on the basis of its assumed purpose (definition 2), but the teleological raison d'être may exist only in the eyes of the beholder. Even more anthropocentric is the view that a system is defined by its being selected for analysis (definition 3). None of these definitions assumes anything about the composition, size, semantics, and goal of the system. Systems hence vary tremendously in their scale and complexity, dependence on other systems, interdependence of subsystems, function, and dynamics (e.g. Simon and Ando 1961; Mesarovic et al. 1970; Houk 1980b; Bhalla and Iyengar 1999).
The existence of systems. Systems may be either natural kinds or "artefacts. In many cases they are both. Consider memory research. 'Non"declarative memory' is a heuristic term referring to an artefactual system; it is highly unlikely that all memory faculties included under the umbrella of this term indeed comprise a natural system. The hippocampal system (see "hippocampus) is a reasonable candidate for a natural system but its boundaries are not known for sure. The status of the "amygdala as a natural system is unclear. And is the cyclic adenosine monophosphate cascade ("intracellu-lar signal transduction cascades) a natural kind or an artefact of erroneous "taxonomy?
The boundaries ofsystems. Systems impermeable to the rest of the universe are termed 'closed'. Those with permeable boundaries are 'open'. In open systems, the variables determined by causes extrinsic to the system are 'input', whereas those dependent upon the action of the system are 'output'. In reality absolute impermeability is nonexistent, but the level of input and output may be extremely low. Open systems differ in their openness. Living systems are open but semipermeable, i.e. they allow some inputs but not others. For example, a neuron is encircled by a semipermeable membrane that allows only selective flux of materials ("ion channel, "receptor).
The relation of systems to other systems. From the aforementioned description it becomes evident that considering any system as truly independent of other systems is an illusion. Systems always harbour other systems and are themselves parts of still others. A paranoiac conclusion is that systems are everywhere, and that everything is a system (for additional notes on the intractability of systems, see Gall 1986). The question arises, therefore, can we deal satisfactorily with variables in a system irrespective of variables in other related systems ("control)? Luckily, the prerogative of a scientist is to decide what 'satisfactorily is in any given system, so that practically we may decide that interactions below a certain level are ignored. The assumption that variables outside the system are held for all practical means constant (and hence are irrelevant) is referred to as the ceteris paribus assumption (Latin for
'other things being equal'). The usefulness of certain "simple systems derives from a focus on one or a few variables combined with adaptation of the ceteris paribus assumption.
The generality of systems. Systems chosen for analysis differ in their claim for generality. Some are 'types', others 'tokens'. Type is a class, or a "taxonomic entity, possibly an abstraction only, characterized by sameness in certain "dimensions. Tokens are the particulars that instantiate the type. For example, in the previous sentence as well as in this present one, the word 'the' appears twice, but it is only one word. Hence, those were more than once, tokens of the type 'the'.1 The distinction between types and tokens depends on the "level of analysis and discourse. "Glutamatergic receptors are tokens of the type '"receptor', but N-methyl-D-aspartate (NMDA) receptors (NMDAR) are tokens of the type 'glutamater-gic receptors', and copies of the NMDAR are tokens of the type NMDAR. Usually the selection of a system for investigation (or of "models) is driven by a wish to understand the type by studying the token. For example, selection of "Aplysia as a model system for the analysis of learning and memory, was not guided by an irresistible urge to comprehend the mental life of a slug, but rather by the realization that it provides convenient advantages that might allow understanding of types of simple learning. In general, defining the understanding of a type as the immediate objective of the research programme is mostly impractical, for the research itself always boils down to the analysis of tokens. But if we were to conclude that no tokens could illuminate anything about the (hypothetical) corresponding types, many investigators would have lost interest in their systems.
Information and systems. Systems encode information, although in our current understanding of brain systems we rarely understand for sure what the specific information is. It is pertinent to note that 'information' has different meanings in different treatments. In everyday language, 'information' refers to knowledge, i.e. the meaning and significance ('semantics') of input and output. In contrast, in information theory, 'information' is a mathematical abstraction that refers to the uncertainty in coding and transmitting data, irrespective of the semantics (Shannon and Weaver 1949; Wiener 1961; Pierce 1961; "stimulus). This efficacy of transfer is measured in 'bits'. One bit (binary digit) is the choice between two equal likely possibilities. Selection of one of four alternatives requires two bits, and so on; the more alternatives, the more bits are needed to select among them. The number of bits required to select among n alternatives is log^ n. Even if the number of bits is known, it tells us nothing about what these bits signify to neither the sender nor the receiver. A major goal of the neurosciences is to decipher not only the 'information' processed by a nervous system, but also to unveil its meaning ("capacity, "internal representation).
Mental and physical systems. Memory, at least in its more complex manifestations, connotes mental activity. In discussing memory systems it is therefore pertinent to ask what distinguishes mental from physical systems. By posing this question one does not, of course, claim that mental systems are not physical; the issue is merely what turns 'mental' mental. A traditional criterion is that mental systems display 'intentionality', whereas physical systems do not (Brentano 1874). In philosophy, 'intentionality' is 'aboutness'; mental systems exist in states ('intentional states') which are about something, for example, belief, hope, etc. In fact, in most of the systems studied in the biology of memory, the distinction between the mental and the physical is not an acute issue, especially if the system in question is simple and the experimental approach highly "reductive. However, when one approaches issues such as complex "declarative representations, "planning, and possibly even highly developed capabilities of "observational learning, the issue becomes relevant, although not necessarily solvable.
Selection of a system. So how should one select a system for the investigation of memory? Idiosyncratic "bias and training background notwithstanding, several considerations are still noteworthy. First, it is useful to choose a system that provides optimal access to the "methods and level of analysis that one wishes to pursue. In other words, as in many other facets of life, the trick is to match aspiration, capability, and availability all together. Second, although one may wish to illuminate a type, in practice a useful token must be selected, which allows fragmenting the problem into approachable segments. Further, the ceteris paribus prerogative should be exploited liberally, yet without being too serious about its validity. And third, never lure yourself to think that you know everything about your system. At the end of the day, sometimes even in its beginning, it will tear off its disguise and present itself as a metasystem or a subsystem of yet another unfamiliar system.
Selected associations: Anthropomorphism, Bias, Generalization, Model, Subject
1The distinction between 'type' and 'token' goes back to the physicist and philosopher Peirce in the nineteenth century, but itself is a token manifestation of a much older type distinction between 'universals' and 'particulars'; see Armstrong 1989; also on 'realists' vs. 'nominalists' in *Ockham's razor.
Was this article helpful?