Biomedical Engineering Reference
In-Depth Information
The outputs of this system, intentional behaviors, are a form of communica-
tion (234). These outputs also represent a means for modulating sensory inputs
to the brain, as distinct from top-down adaptation of input once it is in modality-
specific processing streams (97). The reference to communication theory (234)
is given because it helped foster the revolt against behaviorism. This theory pre-
sented a schema for understanding communication in its most general sense,
namely, how one mind or machine affects another. In it, Claude Shannon and
colleagues focused on the technical constraints on communication, and did not
address the "semantic problem" or the "effectiveness problem" of communica-
tion. Ideas from communication theory were integrated with neural systems bi-
ology only recently in domains such as: sensory representation and memory
(56,58,123,203), reward prediction (229), serial response learning and novelty
assessment (23), conditional probability computation (33), and nonlinear dy-
namics underpinning decision making (95,96). Recently, attempts have been
made to address the "semantic problem" posed by inter-organism communica-
tion, and what constitutes meaning for a biological system. Two viable hypothe-
ses have been advanced, which could be considered as two sides of the same
coin. One hypothesis connects meaning in biological systems to the intersection
of intentional behaviors between organisms (95). The alternative hypothesis
places meaning within the context of organism optimization of fitness over time
and tissue metabolic needs (33). For the latter hypothesis, communication be-
tween organisms utilizes message sets defined by genomic and epigenomic con-
trol of the bioenergetics of metabolism (32).
A number of the general operations and processes of the MIT model (Figure
2a) have been the target of experimental dissection. For instance, when an ani-
mal seeks and finds an object with motivational salience, a set of hypothetical
informational subprocesses appear to be active (Figure 3) (15,33,66,76,96,114,
116,237,138,268). A partial listing of these subprocesses includes the following:
(1) reception of input from the environment or internal milieu across multiple
channels, (2) representation of this input by transient neuronal activity, (3)
evaluation of input representations for sensory modality-specific characteristics
such as color and motion, (4) combination of these representations across mo-
dality at theoretical convergence zones as potential percepts, (5) encoding
of representations into memory and contrast with other stimulus memories, and
(6) evaluation of representations for features (rate, delay, intensity, amount,
category) that are important for the organization of behavior. Feature evalua-
tion encompasses: (a) categorical identification of putative "rewards" or aversive
stimuli, (b) extraction of rate and delay information from the object of worth,
and (c) valuation of goal-object intensity (i.e., strength) and amount in the con-
text of potential hedonic deficit states. The second of these feature evaluation
subprocesses allows computation of a rate function to model temporal behav-
ior (99), and of a probability function for possible outcomes (131,254) (Figure
3). The mechanism by which the output of these valuation and probability (i.e.,
