Explanation-Based learning
<p class=default>
<b>Explanation</b>
<br/>
As scientists, we want to provide explanations for the phenomena we observe. That is, we
want to propose conceptual or concrete systems that can be deemed to be intentionally
isomorphic to (models of) the systems that generate the observed phenomena. In fact, an
explanation is always an intended reproduction or reformulation of a system or phenomenon,
addressed by one observer to another, who must accept it or reject it by admitting or
denying that it is a model of the system or phenomenon to be explained. Accordingly, we
say that a system or a phenomenon has been scientifically explained if a standard observer
accepts that the relations or processes that define it as a system or phenomenon of a
particular class have been intentionally reproduced, conceptually or concretely.
<br/>
Two basic operations must be performed by an observer in any explanation:
<i>(a)</i>
the specification (and distinction thereof) of the system (composite unity) or phenomenon
to be explained; and
<i>(b)</i>
the identification and distinction of the components and the relations between components
that permit the conceptual or concrete reproduction of the system or phenomenon to be
explained. Since these two operations are not independent, when the observer specifies a
system or phenomenon to be explained he or she defines the domain in which it exists and
determines the domain of its possible components and their relations; conversely, when the
observer specifies the actual components and relations that he or she intends to use in
the explanation, he or she determines the domain in which this will be given and in which
the reproduced system will exist. Yet the kind of explanation that an observer accepts
depends on his or her a priori criteria for the validation of his or her statements. Thus
the observer may accept either a
<i>mechanistic</i>
or a
<i>vitalistic</i>
explanation.
<br/>
In a mechanistic explanation, the observer explicitly or implicitly accepts that the
properties of the system to be explained are generated by relations of the components of
the system and are not to be found among the properties of those components. The same
applies to the mechanistic explanation of a phenomenon, in which case the observer
explicitly or implicitly accepts that the characteristics of the phenomenon to be
explained result from the relations of its constitutive processes, and are not to be found
among the characteristics of these processes. Contrariwise, in a vitalistic explanation,
the observer explicitly or implicitly assumes that the properties of the system, or the
characteristics of the phenomenon to be explained, are to be found among the properties or
among the characteristics of at least one of the components or processes that constitute
the system or phenomenon. In a mechanistic explanation the relations between components
are necessary; in a vitalistic explanation they are superfluous. An example of a
mechanistic explanation is: The weight of a body is the sum of the weight of its
components. The relation
<i>sum</i>
, applied to the components as defined by their property weight, determines the property
weight of the body. Example of a vitalistic explanation: Jacques Monod said in
<i>Le Hasard et la Nécessité</i>
(1970) "L'
<i>ultima ratio</i>
de toutes les structures et performances téléonomiques des être vivants est donc enfermée
dans les sequences de radicaux des fibres polipeptidiques, 'embryons' de ces démons de
Maxwell biologiques que vent le protéines globulaires. En un sense très réel c'est à ce
niveau d'organization chimique que gît s'il y a en a un, le secre de la vie [p. 110]."
[The
<i>ultima ratio</i>
of all telenomic structures and functions of living systems is, then, embeded in the
amino acidic sequence of the polypeptide chains that truly constitute embryos of Maxwell's
biological demons that are the globular proteins. It is at this level of chemical
organization that in a very real sense lies, if there is any, the secret of life.] This
statement answers the question -- What kinds of systems are living systems? -- by
reference to the properties of one of their components.
<br/>
In a mechanistic explanation the observer explicitly or implicitly distinguishes between a
system and its components, treating the system and the components as operationally
different kinds of unities that belong to disjoint sets that generate nonintersecting
phenomenic domains. The relation of correspondence between the phenomenal domain generated
by a system and the phenomenal domain generated by its components, which an observer may
assert after enuciating a mechanistic explanation, is, therefore, established by the
observer through his or her independent interactions with the system and with its
components and does not indicate a phenomenal reduction of one domain to another. If it
appears as if there were a phenomenal reduction, it is because in the description all
phenomena are represented in the same domain, and, unless care is taken to preserve it,
the relation established through the observer is lost. The reality described through
mechanistic explanations, then, implies the possibility of an endless generation of
nonintersecting phenomenal domains as a result of the recursive constitution
(organization) of new classes of unities through the recursive novel combinations of
unities already defined. For epistemological reasons, then, mechanistic explanations are
intrinsically nonreductionist.
<br/>
With vitalistic explanations, the situation is the contrary: They do not distinguish
between the phenomenal domain generated by a unity and the phenomenal domain generated by
its components. The reality described through vitalistic explanations is, necessarily, a
reality of a finite number of phenomenal domains. For epistemological reasons, then,
vitalistic explanations are intrinsically reductionist.
</p><p>
</p>