INFORMATION HANDLING IN THE NERVOUS SYSTEM:

AN ANALOGY TO MOLECULAR-GENETIC CODER-

DECODER MECHANISMS

BERNARD L. STREHLER*

Pattern recognition and appropriate response appear to be the common

elements in all adaptive behavior ofliving systems. Ifthe responses elicited
by a patterned input impinging upon such a system result in an increase
in the probability that the responding organism will transmit its genes
to descendants and they to theirs et sequitur, those genes and gene com-
binations which enable their carriers to meet these response requirements
most adequately will tend, in competitive natural environments, to be
selectively perpetuated—a restatement of the Darwinian evolutionary
dogma.
The nature ofthe patterned input to a system is by itselfnot the relevant
parameter. Rather, the relationship ofinput patterns to possible alternative
responses is the key element in such evolutionary interactions between
organism and environment. This relationship is usually incorporated into
the genome of a living system in the form of some kind of complemen-
tarity between an internal pattern1 (structure) and the patterns imposed
* Department of Biological Sciences, University of Southern California, Los Angeles, California
90007. 1 wish to thank Professors Jacek Szafran, Bernard C. Abbott, John Petruska, and Dr. Edwin
Perkins for perceptive criticisms and comments during the preparation of this manuscript. A portion
ofthis work was carried out with support derived from National Science Foundation grant GB-7493
and Atomic Energy Commission contract AT(n-i)-ii3 PA 20.
1 A pattern is defined as an improbable regularity, either in the arrangement of objects or in the
time order of events. Because of the natural decrease in order which results fiom the operation of
thermodynamic principles, as time passes, pattern (order) will decrease in any closed system unless
potential energy is expended to generate or maintain such patterns within a system. The presence
ofpattern is determined experimentally by comparing the linear distances between elements (objects)
or the time distances between events. Unrigorously stated, if the measurements obtained exhibit a
higher frequency ofcertain numerical values than of others (that is, if the numbers are not random),
then one may conclude that a pattern exists, spatially or temporally, as the case may be. Most im-
portant, the observation ofany regularity or pattern within a given randomly selected part ofa system
makes it more highly probable that a similar pattern will be encountered in an adjacent part of the

Bernard L. Strehler · Information Handling in the Nervous System

584 Perspectives in Biology and Medicine · Summer 1969
by the environment. Further, such complementarity, by virtue of direct
or indirect interaction between environmental patterns and internal pat-
terns, increases the probability that the system will do work directed to
the "selected" objective, the long-term survival of the line.
Such complementarities are encountered in various forms at all levels
of biological organization and function. At the level of DNA-RNA
function, it is the complementarities between purine and pyrimidine base
structures which determine the kinds of DNA and RNA molecules that
will be synthesized under the influence of the template pattern [l].
At the level of tRNA amino acylation, it is the complementarities
existing among the structures of the tRNA molecules, the acylating en-
zymes, and the appropriate amino acids that regulate the specificity of the
tRNA charging reactions [2]. At the level of enzymatic catalysis, it is
primarily the complementarity between the tertiary structure of the en-
zyme and its substrate that selectively accelerates given reactions [3].
At the level of adaptive enzyme formation (e.g., Operon model and
allosteric modulation of enzymes), it is the complementarity between the
repressor and inducer (or allosteric agent and enzyme) which yields in-
formation that confers a capacity for an adaptive response that furthers
survival [4].
These and many other stereotyped responses of living systems to pat-
terned inputs have been directly incorporated into the genes of all living
systems. However, as sensory receptors evolved (sensory receptors are,
of course, another kind of system in which complementarity between
input and internal patterns elicits a specific output), particularly those
subserving the detection of events at a distance, patterned information,
whose accurate interpretation has survival value, became more and more
available and complex in its form. Consequently, the ordinary trial-and-
error regimen which operates in the genetic fixation of stereotyped re-
sponse became inadequate to provide optimum adaptation (and rate of
adaptation), except in highly restricted environmental niches.
This fact conferred particular survival advantages upon any variants
system in space and/or in the same part ofthe system at an adjacent time. This fact makes it possible,
within limits, to predict the pattern of the future on the basis of the patterns of the present or past.
It is this fact which lies behind adaptive responses; for those responses which are appropriate to a
given environmental pattern at one time are also likely to be appropriate at some later time if that
pattern is encountered.

585
or mutants which possessed a means to adjust their responses to environ-
mental patterns on the basis of the creation of internal structures on a
nondirect genetic basis. Of course, even such capacities are ultimately
genetically programmed if they exist within a living system, but such
nongenetically imprinted responses are particularly effective if they per-
mit adaptation to occur in the absence of specific mutation and selection.
Because, at the organismic level, the chief adaptive responses to en-
vironmental patterns (of challenge or opportunity) consist of some kind
of movement, either toward or away from some environmental object,
it is of great survival value to incorporate into an organism a system
which can increase the probability of appropriate movement in response
to a given set of environmental signals. An additional advantage is in-
cluded if those movements or other responses which were successful in
an individual's past were somehow recorded nongenetically within the
system in such a form as to elicit the same decision or response in the
presence of a similar or identical pattern of events in the environment.
The material basis for such delible, nongenetic pattern imprintings
exists in the nervous systems ofmany highly evolved forms oflife and in
outline even in some simple forms. The operation of highly evolved
nervous systems consists (inter alia) , then, of the selection and implemen-
tation of appropriate responses from among various alternatives on the
basis of complementarities or resemblances among internal patterned
representations (of patterns preexisting within the memory elements,
whether programmed genetically or on the basis ofindividual experience)
and those patterns which arise from the environment at a particular time.
Essentially, this recognition and reaction role of the nervous system is
identical to other varieties of responses to patterned inputs at lower levels
of complexity, as listed above, but the vectors and structures involved
are, of course, grossly different.
The central concept developed in the text that follows is that the principle
(but not the details) ofpattern recognition and response which operates at the
molecular-genetic level is also the basis ofpattern recognition within the nervous
system and the responses it engenders. In particular, it is suggested that the
mode of pattern processing within the CNS is essentially identical, in
principle, to the decoding function of the mRNA-tRNA system.
A corresponding analogy has recently been suggested in embryological
genetics. This analogy proposes that the differential control of protein

Bernard L. Strehler · Information Handling in the Nervous System

586 Perspectives in Biology and Medicine · Summer 1969
synthesis which occurs in different cell types of the same animal during
development is similarly determined or programmed [5]. Thus, the ca-
pacity ofa given type ofcell to recognize and decode only certain messages
(present as different "cell-specific messages") is postulated, in this model,
to be a controlling factor in cellular differentiation. In effect, the genetic-
code "word set" translatable by a given cell type obviously limits synthesis
exclusively to those messages which lie within that "word set" (or lan-
guage). Considerable evidence consistent with this model has accumulated
recently.
In the present model, the concept is developed that the various patterns imping-
ingfrom the environment through the sense organs are coded (either by the sensors
themselves or at a more central station) in theform of unique time sequences of
pulses; and, further, that the "selection" of appropriate responses is achieved
through decoder systems that operate analogously to tRNA function and consist
of single sells which are the individual decoder elements. Provided that such
selective pattern discriminators (decoders) are suitably unique in their
responses, the present model provides a means suited to the selective
channeling, storage, and comparison mechanisms which are essential to
the analysis of incoming patterned signals and to the generation of "pre-
dictions" upon which adaptive reaction may be based.
Some extensions ofthese concepts to more complex pattern processing,
a comparison of the properties of the model with certain observed phe-
nomena, and a series of behavioral properties of the nervous system pre-
dicted by this model are presented. Specific experimental tests of these
predictions are outlined. The striking feature of the model is its essential
simplicity and its analogy to well-established mechanisms which are known
to operate at the molecular-genetic level.
STATEMENT OF MODEL

The essential function ofthe nervous system is the channeling ofsensory

data patterns and the selection and initiation ofa suitable response (usually
a movement), with or without preceding comparison, through some as
yet unknown means, of the sensory-derived patterns with stored pattern
representations. Several means for achieving this analysis (sorting) and
subsequent reaction have been considered in the past.
The traditional and most obvious possibility (essentially an elaboration
of the simple-reflex-arc concept) consists of a network of input neurons

587
which make specific and appropriate connections through internuncial
cells to an efferent (effector) system. Whether or not an output discharge
ofone kind (location) or ofanother occurs is determined in such a model,
by the presence (or absence) of appropriate connections between the
input and output via such intervening neurons. Such a model is clearly
very suitable for the execution of simple, stereotyped reflex responses.
However, this model appears to involve grave, perhaps insurmountable, diffi-
culties when it is applied to higher functions (particularly to conditioned or
learned behavior), for it postulates explicitly the instantaneous establish-
ment of new functional connections between neurons or the facilitation of
existing connections in some unknown, selective, mysterious, and highly
complex fashion.
The imprinting of experience is, in these classical theories, a process
that involves and invokes prodigious complexities and selectivities of a
most subtle and unknown kind. No really satisfactory, widely accepted
mechanism for such imprinting has yet been proposed. A class of obser-
vations which tend to lend plausibility to such a view is the apparent
localization of specialized function in different parts of the cortex and in
subcortical areas. However, the findings of Lashley [6], more recently
substantiated and extended by other workers,2 have cast some doubt on
the general applicability of the concept, because it now appears that the
central representations of sensory input are not so uniquely localized as
was earlier thought to be the case.
For this reason, the present model, which makes use of a concept of
information (pattern) channeling different from the establishment of new
connections, qffers an alternative solution that appears to possess great
advantages over the reflex-arc-type model. In direct contradiction to the
classical model, the imprinting of memory and associations within the CNS as
outlined in the present theory is viewed as a decrease in the number of effective
neuronal connections required for transmission, rather than as the increase of
such connections—as had earlier been expected.
CODING AND DECODING

The essential feature of this model is that the processing of input pat-
terns takes place through the use of coded signals rather than through the
* See excellent discussion of the problem and recent experiments in the article "The Neuro-
physiology of Remembering," by K. H. Pribram, Sei. Amer., 220:73, t9^9-

588 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
aegis of unique connections between afferent and efferent fibers. The
advantage ofcoded signal processing (obviously the mechanism at the molec-
ular-genetic level) lies in the fact that in such a system there is no require-
ment for unique, permanent, and exclusive connections between succes-
sive elements in a reaction chain (either neurons or translator molecules) ;
rather, a relatively small number of signal transmitter cells (or tRNAs)
can serve to transport patterned information from a given source to a
great number of potential receivers. If the response of the receiver is
determined by the signal pattern (i.e., decoding) rather than through unique
individual connections between elements, no necessity exists for establish-
ing or maintaining such unique connections.
This kind of solution is, in fact, employed in telephone networks, for
by virtue of the decoder function of the switching relays (when a given
number is dialed), it is possible to transmit information very selectively
without the cumbersomeness and expense implicit in a separate line be-
tween each telephone in a metropolitan area and every other telephone
in such a service area.
The central question, of course, is not whether such a pattern trans-
mission and recognition system is possible within the CNS but rather
whether the known or reasonably extrapolated properties of neuronal
elements can provide a suitable material-anatomic basis for such coder-
decoder networks. Whether the complexities involved would generate
new problems of such intricacy (as regards unique structural relationships
and instabilities) that evolutionary processes could not operate to provide
workable solutions with realistic probability assignments is a similarly
cogent question. In particular, in order for a coder-decoder system to
operate effectively as a basis for pattern recognition-analysis and subse-
quent action, the basic coder and decoder system elements must be very
simple indeed, and must nevertheless possess the potentiality for great
specificity and stability (reproducibility in function).
These requirements are postulated in this model to reside in individual
cells which act as unique coding entities and in other cells, located else-
where in the system, which act as correspondingly unique decoding ele-
ments. The kind ofcoding envisioned makes use ofa time pattern or display
of separate impulses rather than of a spatial pattern—which constitutes an
alternative system. The latter type of system may be important in the
analysis of visual phenomena.

589
 The four well-established phenomena of (i) finite conduction velocity,
(2) synaptic spatial summation, (3) facilitation, and (4) inhibition are the
sole basic elements employed in the present model, outlined as follows:
In figure 1, assume that a pair of (coder) cells designated as Ac and Be
(located in close association with sensory cells, or more centrally) generate
a train of impulses each time they are stimulated, possessing the time-
patterned display 1 [ I (^4) and I 1 ] (B), respectively, and
that the corresponding decoder cells A¿ and Bd are only responsive (i.e.,
discharged) when the discrete patterns A or B are applied to them,
respectively. Then, if the connections between Ac, B6, Aa, and Bd are as
noted in the figure, the decoder cell, A¿, will only fire when the pattern
IHPUT "A" INPUT INPUT "X"

CODED SIGNALS

T
Fig. i

of signals generated by Ac reaches it through the common transmitter

cells, T1. Similarly, the other cells in the ensemble will only be discharged
when signal patterns appropriate to them impinge on them.
Coder cells.—In this system a single cell is postulated to generate an
appropriate coded signal on the basis of its unique structural properties,
as shown in figure 2. In this view the path lengths (conduction times)
through the various axonic connections which Ae (or Bn etc.) makes are
proportional to the time delay between successive discharges of the "common
transmitter" cell, T1. Note that a single input pulse (to Ae, for example)
produces a succession of three discharges of cell T1, with a time sequence
11I as required. Provided that the effective delay-line char-
acteristics are such as to avoid a succession of impulses on T1 that fall

590
Bernard L. Strehler · Information Handling in the Nervous System
Perspectives in Biology and Medicine · Summer 1969
within the latter's refractory period, there seems to be no obvious objec-
tion to this single-cell coding function. (Alternative multicellular models
are also possible, but will not be treated here.)
Decoder cells.—The same delay-line property of multiple paths of dif-
ferent lengths between two cells could effectively serve as the basis of a
single cell's function as a decoding device (fig. 3). Assume a decoder cell,

""Η

Fig. 2

(via x) (J.
(via y) ____
(via ?) ,
TIME-
and which give the following
pulse sequence onto T,

LL_I i_U
which results in the sequence ' I 1 out of T„.
Fig. 3

Ad, which possesses a relationship inwhich signals impinging on T3 through

paths x, y, and ? are shown in the lower portion of the figure. Provided
that cell T2 requires the spatial summation of impulses arriving simul-
taneously from the alternative routes x, y, and z, that cell (T,) will fire
only when the large pulse, indicated as (}), arrives. If, provided further,
cell T2 is able to be discharged by any arriving pulse for a brief interval
following its original discharge by the coordinate large pulse (f), it follows
that cell Ad mayfunction as a generator ofits own coded signal (codon) and of
a series ofinverted signals (anticodons), as shown in figure 4.

591
 In summary, both coding and decoding, as well as recoding and codon
inversion, can be achieved by single cells (in conjunction with an input cell
for Ad), provided that that output cell, T2, requires spatial summation for its
initial discharge and is brieflyfacilitatedfollowing its initial discharge. In essence,
this simple arrangement permits each T1 cell to be discharged by and to trans-
mit further only that unique signal (codon) which triggers the "decoder
diad" (Ac + T2) followed by a series oftime-inverted copies (anticodons)
of the codon.
-A

I 1 Ki)I 1 L..-
A/
A
Fig. 4.—Note that the output contains a "spurious" pulse. In general, the number of such pulses
will equal the following expression: (n2 — 3« + 2)/2, where « = the number of elements in the
codon complex.

EMBRYOGENY OF CODERS-DECODERS

The physical requirements during embryogeny for the production of

an orderly array of cells which possess progressive changes in codon
generating, recognition, and regenerating abilities are exceedingly simple.
Such (semi-) systematic arrays of decoding elements, according to this
model, are represented by the cortical sensory and motor representation
areas of the brain and would appear to confer the usual advantages im-
plicit in the selective location of biological structures which have related
functions, in proximity to each other, and to ensure, as well, the inclusion
of "all" potential decoders within the system. In keeping with the above,
if, during the embryonic development of the nervous system, there is
differential lateral displacement of the region between the cell bodies of
Ad and T2, as successive axonic connections are established there will
also be established, concurrently, approximately appropriate coder or
decoder arrays, as shown in figure 5. Or, alternatively, if successive con-
nections between two cells are established through axons which possess dif-
ferent diameters (perhaps because of differences in the cell's growth rate
during the period in which successive axonic branches were established),
different delay characteristics for different paths would also ensue, as a

Bernard L. Strehler · Information Handling in the Nervous System

592
Perspectives in Biology and Medicine · Summer 1969
result of the known dependence of impulse transmission rate on axonic
diameter. Further, because the course of development on the left and
right halves of the brain might be expected to be approximate mirror
images spatially, such an orderly process would also generate similar
patterns of coding-decoding (spatial, but not time-base, mirror images)
on the contralateral sides of the brain.

STAGE "A" ETACE "B" STAGE 11C"

DIRECTION x "
OP -, , . , . ¦ "I SLIPPAGE
ZOEE
SLIPPAGE
"*· -
-*-*·->- ^y J "»¦ ·*; ·*· +* N2I * *J ZONE

Fig. s

IMPRINTING OF "new" EXPERIENCES: MEMORY AND ASSOCIATION

The model as developed above furnishes a plausible basis for stereo-
typed simple coded inputs to the decoder cells, such as may exist even
between internuncial cells in a reflex arc, and for the fact that such re-
flexes can be augmented through repetition. The output cell, T2, is in this case
connected to an appropriate effector (motor) system.3 But the nervous
system of higher organisms handles complex patterns of inputs, both in
time and in space, as well as simple-reflex-initiating inputs.
This capacity to handle complex inputs generally requires that signals
(coded in this model) be combined in some fashion so as to generate an
appropriate response to a complex input pattern. More crucially, if coded
patterns are to be combined in some fashion, it is apparent that certain
decoders must, under appropriate conditions, respond to mixed inputs from
several different sources! Because it is unlikely that sufficiently fine dis-
crimination will be afforded by a pair of pulses, at least a time-sequence
triplet (!) as illustrated would seem to be necessary.
It also appears probable, for purposes of combining (associational)
coded inputs, that certain such complex decoders may require as many as
six time-coherent pulses in order for them to become activated. However,
3 It is important to keep in mind that there is no apparent selective advantage to the production
of coded signals in outputs to motor units. Coding and decoding are most likely to have utility in
the sorting out or selection ofappropriate responses, not in their motor implementation.

593
although a complex of six or so coordinate pulses would be required for
an initial activation of an "associative decoder," it is also evident that a
substantial evolutionary selective advantage would accrue (because the
capacity for associative learning would thereby result) if the original
discharge of a particular decoder by the appropriate pulse complex were
to be followed by an increase in the probability that less than six, and
eventually as few as three, appropriate pulses would become sufficient
to discharge such associative decoders (fig. 6).
INPUT "A"INPUT "B"
I I I COMBINED INPUTS (A+B | | | A B

INPUTS: R 1111
T
ASSOCIATIVE
DECODER
s Mil
* 1 I I U
R STUVW
?-
?
?
ASSOCIATIVE
TRANSMITTER
w

SUMMATED
INPUT 1 H Il 1
ASSOCIATED
OUTPUT I ??????: il I

INVERTED
OUTPUT JJ
(anticodons)
-B-A
Fig. 6

For these reasons (as shown in fig. 7) and provided that the number oj
simultaneous pulses arriving through the associational decoder (Ada) at
Ta needed for the discharge of that cell (T0) decreases if (Ta) has been re-
peatedly discharged, such change in spatial summation threshold will per-
mit the emission of an associative output in response to a single input codon
which is patterned after the original DOUBLE codon input!
An increase in the intimacy or effectiveness of the connections between
decoder and Ta will thus serve as a basis for associative memory, condi-
tioning, and learning. Moreover, if a similar increase in the probability

Bernard L. Strehler · Information Handling in the Nervous System

594
Perspectives \n ßichgy and Medicine · Summer 1969
that three time-coherent pulses will fire a T2 neuron occurs, even in non-
associative decoders, there exists a basis for the selective transmission of
such a codon (with a higher probability than existed initially) if such a
decoder had previously been caused to fire repeatedly. In other words,
repeated inputs of the same codon will result in a lowering of its rejection
threshold, provided that the firing of T2 is facilitated as a consequence of
its prior activity. Thus, a reasonable basis for nonassociative recall or non-
conscious recognition of previous inputs exists in this model. To enlarge
on this point, ifthefiltering network (which is essentially what such a system
represents) is relatively resistant to decoding on an initial encounter with a
INPUT "A" OR INPUT "B"
INPUTS I I I OR
(A)~ (B)
GIVE

f OUTPUT: M lili (-A-B)

-B-A
-A

I Il I
^v (OVERLAPPING -A-B)

Fig. 7.—In the lower, conditioned, case, note that, although the "codon" A J___I____I alone
(or, alternatively, B alone) will cause discharge of T„, the output will consist ofboth the —A and the
— B codons.

signal, but becomes more sensitive with usage, this network will tend to trans-
mit selectively patterns which resemble past patterns and will also tend to
reject new ones somewhat selectively. This is analogous to the augmenta-
tion of a reflex pathway through repeated usage—that is, memory.
The key concept presented in this section is that learning and associa-
tion processes are not to be considered as dependent on the establishment
oí new neuronal connections, but rather as the sensitization (or simplifica-
tion of the conditions under which a given input pattern will elicit the
response) ofa cell to which the decoder-recoder cell makes its preexisting
connections. On this remarkably elementary basis, both simple memory and
associative memory can be explained in terms of fundamental, specific, and
established properties ofneurons.

595
 DELIBLE MEMORY, THE REFLECTED PATTERN, MULTIPLE
PATTERN SEQUENCING, AND RECOGNITION

The essential feature of the sytem, as thus far described, is the selective
transmission of certain patterns of time-coded signals (codons) and cor-
responding anticodons, and the selective rejection ofothers. Such selective
transmission of different patterns, coded by central stations involved in
sensory pattern transmission to the decoder centers (or by sensory re-
ceptors themselves), may serve as the basis for an immediate motor
response. This will occur if the output cell, T2, is directly or indirectly
connected to the motor system. But more importantly, selective codon
transmissionfurnishes the basis for thefurther processing of the complex output
of T2 by other neuronal elements ofthe CNS. Such filtered, patterned output
of memory and association "banks" may, for example, be transmitted
to other decoder systems existing elsewhere in the CNS, such as those
elements which are involved in "conscious" behavior and "awareness."
DELIBLE MEMORY IMPRINTING

A well-established feature of the nervous system is its ability not only

to be imprinted by patterns which arrive at about the same time but, even
more importantly, its ability to form associations between impulses arriving
from elements of the same or of different sensor systems at different times.
In order that such complex interactions (comparisons) of inputs at suc-
cessive time intervals take place, it appears likely that some system is present
which is capable oj retaining ordered, complex pattern inputfor some brief time
interval following its occurrence—a transient, or "delible," memory, as it
were.4

Ifcomplex input patterns are displayed for a reasonable half-life in such

a delible memory bank, independent of previous learning, it would be
possible for the patterns passed through the memory banks to interact
with briefly persistent patterns imprinted on a delible display in such a
way that pattern "recognition" and reinforcement could ensue. In other
words, recognition involves the comparison of "current" input patterns (and
sequences of patterns) with patterns imprinted in the memory at some earlier
time.

4 The simple experiment ofrecalling the patterns ofwords or of sights which have just occurred
indicates the reality of such temporary imprinting and retention for several seconds following
sensory input to such a time-display system or systems.

596 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
 In order for such a delible display to persist, some means oftemporarily
"imprinting" the continuously arriving sensory inputs in an order closely
corresponding to their order of arrival is needed.
The properties of such a delible memory and display system should
be not unlike a cylindrically shaped cathode-ray tube on which a phos-
phorescent image persists for slightly less than a full sweep of the surface.
Such a screen is diagrammed in figure 8.
Such a system should possess the ability to retain the input in essentially
undistorted form and to project this record at a later time so that com-
parisons between the delible pattern elements and the patterns reflected (i.e.,
those returning) from the memory banks (cortex) may take place with high
DIRECTION OF SWEEP

CATHODE
LOCATION OF
"PRESENT" SWEEP

FAOING IMAGE OF
PAST PATTERN INPUTS
Fig. 8

precision. One possible solution to this problem would be a system re-

sembling that shown in simplified form in figure 9.
This delible memory device (which should not at all be considered
as a literal model) operates as follows: Input signals (IS) are distributed
to all of the image retention elements (IRE) simultaneously. However,
because the initial discharge of these IRE elements requires that they
receive simultaneous inputs from both the IS sources and the primary
sequencer (PS), any given IRE will not fire in response to an input pulse
unless the sweep pulses from the PS are in the appropriate portion of
their cycle. If, in the following period (one-twentieth of a second to five
seconds), the IRE elements are able to be fired by any of the signals
emanating from several scanning sequencers (SS), it follows that repetitive
patterns of coded output will be emitted as long as the IRE elements remain in
thispoised orfacilitated condition. As the elements lose their ability to respond

597
to the scanning sequencers, pattern production will cease to be emitted at
the output. Note that the operation of several scanners in tandem confeis
upon the system variable output delay and "repeat" characteristics.

PRIMARY SEQUENCER IMAGE

(SLOW period) ^_^. RETENTION
\Z^SELEMENTS

INPUT

LEGEND
: REQUIRES SCANNING SEQUENCERS (SAME
SPATIAL SUMMATION SWEEP RATE BUT DIFFERENT
- DIRECT TRANSMISSION PHASING - FRAMESHIFTÎ)
- FACILITATED TRANSMISSION

Fig. 9

recognition: COMPARISON OF PRESENT AND PAST INPUT

The "recognition" of codons recorded in a delible memory system

(i.e., a determination of their identity with or lack of identity with
patterns transmitted from and through the memory and associational
banks) requires that an output pulse series (the recognized codon itself?)
be transmitted to a "recognition detector system." A simple means of
achieving this consists of a series of time-delayed inputs to a detector cell
in which simultaneity between the impulses arrivingfrom the permanent memory
andfrom the delible memory is the condition prerequisite for further transmis-
sion to the recognition detector.
However, as is apparent from a casual inspection of figure io, a crucial
condition for the passage of the codon, A, through this recognition de-

598 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
tector is the simultaneity ofarrival (i.e., the phase relationship between the
two separate pulse trains) of the separate pulses making up a codon and
originating from two separate sources. In solving this problem, the delay-
line conceptemployed in coder-decoder operation outlined earlier maybe used
as a means ofadjusting the phase relationships between two signal trains. Thus, if
the two inputs are applied to a series ofdetectors which differ ina systematic
fashion in the time delays between the separate inputs, one of these de-
tectors may fortuitously detect an identity between parts of the two
different pulse patterns arriving by separate paths. This is shown in fig-
ure II.

INPUT FROM

O PERMANENT
MEMORY.

ILUlIi-L(N)
JLiLiLL ILL (M)
SUMMATION
INPUT
INPUT M
FROM
DELIBLE
MEMORY
o- _<( DETECT« CELL
OUTPUT A
? ? ?

Fig. io

Note that any simultaneous pulses arriving from M and N will trigger
the discharge of T3 (simultaneity detectors) and T4 (the collector neuron).
Should the accidental "noise" transmitted be significant, it may be reduced
by the arrangement in figure 12, in which the T4 system is facilitated only
when several "simultaneity pulses" arrive within a sufficiently short time
interval to have a high probability of constituting a "true" codon.
In this event, the signals arriving at T3, as a result of a single codon
triplet A, Il I , will be

but the early triplet, Ae, arriving by pathway G, causes the codon gate
cell to open for a briefperiod (approximately one codon transit time) and

599
to transmit the "late" pattern, Al, to cell T4. In this way, spurious pulses
(except those occurring during the Al period) will be eliminated unless
three such random pulses arrive within a suitably short interval and acci-
dentally open their gate cell.
SIGNAL PHASING IN ASSOCIATION

The problem ofphase manipulation (positioning) is crucially important

in pattern recognition; but phasing is equally important in associational
imprinting, because there is a serious restriction if associational decoders must
recognize two incoming codons in their original specific phase relationship.
Therefore, there is a substantial advantage to mixing such pairs of codons
INPUT M INPUT N
SYSTEMATIC
DELAY
LINE
r PHASERS

ONLY

J
[PULSES ARRIVING
SIMULTANEOUSLY
WILL FIRE T, -?-

??? PULSE FIRES T. ("COLLECTOR")

OUTPUT

Fig. i i

©
TEMPORARILY FACILITATING GATE (PATHWAY G)
DELAYED SIGNAL
TRANSMITTER
CODON GATE
(PATHWAY H)

Fig. ia

600 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
(or anticodons) in all possible phase relationships so that they may act to
trigger one ofthe appropriate associational decoders. This can be achieved,
and the purity of signal maintained as well, if the inverted output of
decoder elements is able to trigger an anticodon decoder cell, a cell
identical with the codon decoder except for an inverted pattern sensitivity.
(Also note identity with coder configuration!) Obviously, such an anti-
codon decoder will remit multiple copies ofthe codon following its facilitation by
the anticodon sequence's action. Thus:
The decoder detects the codon [A = Il I ) and emits A followed by a series
of (1 II. I II·) anticodons (—A, —A, etc.) in various phase relation-
ships.
The anticodon decoder detects the anticodon (-A= I II) and emits —A
followed by a series of codons {A — Il 1 . Il I . etc.) in various phase
relationships.
In this fashion, multiple copies of A (and of —A) will be generated
and give rise to self-sustaining reverberation or "ringing" within the
system unless they are suppressed periodically. These copies will occupy
random phase relationships with each other and with other, like or unlike
patterns originating elsewhere, because the inverted copies are triggered by ran-
domly arriving pulses. Such phase randomizations would be expected both
to facilitate the imprinting of associations and to provide multiply phased
copies of memory-bank outputs suitable for the "recognition" features
outlined in preceding paragraphs.
To speculate further on the possible details ofsuch circuits in the absence
of direct evidence for their existence is probably an empty exercise. But
the value of the model, if any, is in its demonstration that pattern se-
quencing, pattern phasing, and complex pattern recognition are possible,
in theory at least, as a direct result ofknown properties and interactions of
neurons (spatial summation, facilitation), plus phenomena of which the
reader can readily make himself aware through simple tests of his own
image (pattern) persistence, retention, and filtering abilities.
At this point it is also relevant to emphasize that the scanning circuits
discussed in the above paragraphs have essential qualities required for the
existence of the phenomenon of conscious awareness, for a central feature of
conscious behavior is the reflexive comparison of the correspondence between
impinging sensory pattern sequences and like patterns that are stored in the
memory "bank." These interactions, in a real or operational sense, can be
considered to constitute "self."

601
 STORED PATTERN COMPARISON: REASONING AND PREDICTION

Without the postulation ofadditional properties, or complexities, of the

model system, the features here enumerated are also capable ofaccounting,
in principle at least, for the most highly evolved neural processes (in
man)—the ability to reason, that is, to predict relationships on the basis ofa
comparison of several independent multiple codon pattern representations. Essen-
tially, reasoning consists of the synthesis of new patterns from combina-
tions of old ones.

A U___I
» LU
c U__I

PATTERN DOUBLET A-B| | I I I I

As //\
PATTERN DOUBLET B-CI I I I I [
^AxBv.Cx

The synthetic triple pattern, | [ I I I I I I I

TIME ----*¦

represents the construction of an association between A and C-

Fig. 13.—This process is very analogous to the methods that molecular biologists have success-
fully employed in reconstructing (i.e., predicting or deducing) the sequence of amino acids in pro-
teins and the nucleotide sequences in the tRNAs. The basis of these biochemical deductions is the
piecing together of different fragments derived from the same kind of molecules through several
very specific fragmentations of the original molecules involved. The assumption is that the arrange-
ment can be deduced from the relationships between partial fragments derived from separate degrada-
tive reactions. Similar principles are also used in gene mapping.

This ability, so uniquely useful to man in his domination of this globe,

differs from the properties thus far discussed, in that reasoning involves the
conscious recognition ofprobable associations between two sets of multiple pat-
terns, some ofwhose elements have not yet been experienced as a combined input.
To illustrate, iftwo separate coded input patterns, A and B, are frequently
found in close temporal or spatial association with each other and if,
similarly, B and C are also found in frequent (i.e., probable) association,
then a high probability also exists that patterns A and C will be found in
association. This is diagrammed in figure 13 for two such pattern doublets,
A-B and B-C, and for the deductive (predicted) doublet A-C.

602 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
 The deduced relationship between A and C, possible through the as-
sociative recall mechanisms and properties of the model discussed in an
earlier section, together with the complex-pattern comparison device de-
scribed in the immediately preceding section ofthis paper, will result from
the associated projection of the ^4-C-pair code sequences as a result of the
nearly simultaneous reflection ofthese codons from their respective niches
in the associative cortex (in association with B). Following this, their
reinsertion into an output pattern from the recognition center makes
them available for such decoders (and memory representations) as may be
triggered by the new synthetic, deduced output pair, A-C.
Reason and prediction (which essentially consist of the assignment of a
greater than random probability to the expectation that several patterns
which have not, up to a given time, been observed in direct association through
sensory inputs will, in fact, befound to be associated in the external [or internal]
world if appropriate measurements are made) have an obvious survival
value on which this species, in particular, has capitalized thus far. A close
resemblance to the principles of the scientific method is also evident.
CHANNELIZATION OF CODING PATTERNS: ATTENTION, REPRESSION,
AND BRAIN WAVES

Insightful and objective study makes it evident that not all of the input
signals arriving from the sensors receive equal attention or are processed
equally. Essentially, attention tends to be centered on those aspects of
input (i) that are highly predominant, (2) which change with time, or
(3 ) in which change versus stored impression is large.
This focusing ofattention appears to confer survival advantages because
the data that are of highest probable significance in food acquisition, defense,
and reproduction are those which signal (1) a very large or (2) active en-
vironmental object (predator) or (3) a difference between a recollected pattern
and a pattern observed in the present environment.
Repetitive patterning from the environment tends to divert attention
from such a repeating sequence, perhaps because of some kind of fatigue
of the reflected patterning elements (habituation). On the other hand,
any new pattern elements imposed appear to focus attention on these
elements and to elicit a concurrent search among storage elements for
their "meaning," that is, the relationship of stored memories and associa-
tions to such dominant or attention-focusing patterns. We seem, in fact,

603
to be unable to concentrate fully on more than one environmental or
internal pattern complex at a time. Marked changes in a pattern derived
from a previously «onattention-holding modality will, however, quickly
divert attention from a previously acutely monitored pattern.
These observations suggest that a given sensory modality, which has
preeminence because of its striking or changing nature, maintains, by
virtue of its operation, a selective advantage over other pattern recogni-
tion and processing centers. Further, the heightened acuity which a given
attention-commanding modality generates for like classes of patterns
suggests that some mechanism exists for the selective accentuation or
repression of those areas of the associative or memory banks which are
appropriate to or inappropriate to a given situation.
One mechanism through which such selective attention might be gen-
erated is via an automatic depression ofthe sensitivity ofcenters which are
"currently oflesser probable significance." This may be achieved through
a continual sweeping clean, as it were (by repressive signals), ofthose areas
which are not in prime use. Such repressive signals, in contrast to those
employed up to this point in this model, would decrease the sensitivity of
decoder systems to current input and, in addition, repress reverberation or ringing
between codon and anticodon patterns. Their imposition could be modu-
lated through tracts of inhibitory fibers.
If such repressor signals (which are perhaps detectable as the so-called
brain waves) were themselves modulated in their intensity in inverse proportion
to the number of patterns transmitted by any given modality, then unused or
underused modalities would themselves become repressed! The decrease in certain
brain wave activities when attention is focused on specific modalities may
represent expressions of such selective de-inhibition, or modality channel-
ing [7]. Such channeling might also confer advantages through the pre-
vention or reduction of signal "spillover" from the dominant (i.e.,
"important") inputs into "nonrelevant" analysis centers.
PATTERN SIMPLIFICATION (ABSTRACTION), CODED REPRESENTATION
(SECONDARY SYMBOLIZATIOn), AND LANGUAGE
The efficiency and specificity ofpattern handling proposed in this model
confer a uniqueness to signals which permits their analysis and channeling
with dispatch. By analogy, it might also be expected that there would be a
substantial advantage in pattern processing if the "sense" or meaning of

604 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
the primary patterns perceived were to be secondarily recoded (abstracted)
in some more readily accessible or usable form. Such an abstracted form of
informational pattern, devoid ofthe irregularities and extraneous elements
in the input, should still permit the (uncritical) essence of object, action,
and relationship to be handled at a level of accuracy sufficient to permit
effective prediction and response, based on such "abstracted" input.
The process ofcoding per se constitutes an abstractional event, provided
that the coded pattern produced possesses less complexity or variability
than does the input; but once the essence of a pattern (i.e., the general
group into which it falls) is recognized, there is an enormous advantage
to its simplification, for further transactional purposes, into a form that
requires no repeated critical reappraisal.
Among our ancestors, a model for such simplification or abstraction
was developed as "language." Certain emitted combinations of sounds
(possibly of evolutionary value initially as devices suitable to communi-
cate simple and recurring relationships, opportunities, dangers, or desires)
were perhaps always or frequently associated during the learning process
with other sensory signals arriving from the outside.
Because of the nature of associative learning, postulated above, such
sound patterns would themselves become part of the output from the
cortical recognition and association centers and would constitute a part of
the enlarged associated pattern elicited from the associative cortex, even
though they were not themselves part of a particular subsequent sensory
input pattern. Such internal-associated representations of sound patterns
encountered at an earlier time and later evoked by nonauditory stimuli
would be expected to be channeled into the auditory recognition centers
and to be handled there essentially as would be "bona fide" auditory
signals from the sensors. It is interesting to note the extent to which we
translate (recode) visual impressions into verbal representations, even in
our private worlds (i.e., while musing).
Because the auditory arrays of patterns have specific associational
"meanings" which can be regenerated when the abstracted sound repre-
sentations are generated, a basis exists for the symbolic (i.e., abstracted)
treatment of the auditionmeanings independent of the detailed arrays of
patterns these sounds symbolize.
Thus, the objects, actions, and relationships which words convey are
simplified, abstracted patterns that correspond to recurring objects, events,

605
and relationships; and actual events and recollections can be (and are)
translated into words (coding) which can be further manipulated without
continual reference to the enormously more complex evoking patterns
which arrive from the environment.
The advantage implicit in this symbolization and simplification system
is enormous, for it permits the use of a few hundred or thousand (rather
than many millions) such stereotyped patterns (sound sequences, i.e.,
words) to perform most of the necessary data processing; it, in effect,
reduces the number of conflicting choices that may inhere in a detailed
analysis of raw input signals.
Man may owe his preeminence in the biological world not so much to
the fact that he can communicate effectively with other members of his
species, but rather to the fact that he can arrive at the vast majority ofhis
decisions wisely and rapidly through the analysis of symbols (a simplified
secondary recoding) rather than through analysis ofliteral input patterns.
There is, however, an implicit source of error in such symbolism and
simplification, for success through such devices in the analysis ofrelatively
simple and recurring problems does not necessarily carry with it a pre-
diction that unique or complex input patterns will be effectively or ac-
curately processed. Semantic errors (i.e., taking literally the word for the
object) are a source of much misunderstanding, confusion, and even con-
flict among people. Senses mislead much less frequently than do words, for
every time one uses an abstracted symbol in a logical transaction, one
takes the chance that the symbol (word) and referent (object or event)
differ from each other in some important way [8].
AN INITIAL CRITIQUE AND SOME TESTABLE PREDICTIONS

It would be unwise at this time to attempt a searching critique of the

ideas presented here, for one of the potential values its presentation
affords is a dissective criticism by those who are more schooled and
skilled in such matters that I am. However, the model does not appear
to be at gross variance with the major facts ofneurophysiology, certainly
not with those of which I am aware.
Such strengths as the essential model or its components may possess
lie in the elementary simplicity of the basic coding and decoding mecha-
nisms outlined and in the fact that such coding and pattern perpetuation
(memory) and extension (association) mechansims as are suggested are
based upon the most elementary of neural phenomena. As pointed out,

606 Bernard L, Strehler · Information Handling in the Nervous System

Perspectives in Biclogy and Medicine · Summer 1969
the model also does not require extraordinarily complex developmental-
genetic specifications or "quantum jumps" in mechanisms in order for a
real system ofthis nature to evolve progressively from the relatively more
simple pattern-processing mechanisms present in lower forms of life to
the more highly evolved ones that human beings employ.
A number of the more perplexing findings and questions which have
arisen during the last three decades are highly amenable to explanation
through this model. Perhaps the most central of these is the series of
observations by Lashley that, despite the localization of certain functions
within the cortex (e.g., motor area) in a highly ordered array or "map,"
many other functions are, at the very least, highly diffusely represented in
the cortex. For this reason the partial destruction ofany particular area of
the brain may lead to surprisingly small and subtle behavioral disturbances.
These results would be expected to follow if the memory and associative
and logical operations were not spatially represented in a single domain
but rather were more widely scattered among various similar or over-
lapping parts of the brain, and if alternative routings of sensory and ana-
lytical inputs (such as those implicit in the present model) were feasible.
A second major observation that is amenable to this model is the finding
that relearning may occur—relearning in which parts of the cortex may
become engaged which were not originally the main areas through which
such lost or damaged functions were channeled.
A third finding that may be relevant is the observation that amnesia,
such as may be induced by a blow on the head, frequently involves the
initial recall of time-distant events rather than recent events. If early rec-
ollections were to involve decoders which have become more fixed
(i.e., decoder elements which require fewer simultaneous pulses to sum-
mate effectively in the recognition and recall processes) than recently en-
coded memories require, it would be expected that trauma-induced de-
formations in decoder-transmitter structures would have a lower proba-
bility of erasing highly reinforced (older) memory circuits than would
those which represent more recent acquisitions of stored patterns.
Fourth is the suggestive evidence presented by Roeder [9] on the avoid-
ance reflex exhibited by moths to the ultrasound frequencies emitted by
bats' sonar devices (fig. 14). The record shown below suggests that the
avoidance signal generated by the sensors is a patterned series oftime-spaced
discharges originating in the sensory neurons involved.

607
 Fifth is the fact that impulses derived from the two different cerebral
hemispheres, as shown by Sperry [io], are nevertheless capable (despite
massive damage to or obliteration ofthe corpus callosum) offinding their
way into the opposite hemisphere and even, in some cases, ofbeing handled
there efficiently. Such massive malformations and damage to conduction
tracts produce surprisingly subtle or almost undetectable behavioral
changes in many instances.
Sixth is the fact that this model may well account for the remarkable
ability of transplanted amphibian eyes (i.e., exchanges or inversions) to

Fig. 14.—Oscillograph record of the pattern of pulses transmitted (lower tracing) by the ultra-
sonic acoustic detectors of a moth in response to the bat "cry" (upper tracing). Note particularly
that there is a "precise" time delay between the three pulses making up a train (codon?) and that
the production of the second and third pulses is repressed if the first fails to appear, indicating time
interdependence in the discharge of these three sensors and their axons (after Roeder [9]).

establish normal functional connections to the cortex, a phenomenon also

ingeniously explored by Sperry [ii] and more recently byJacobson [12J.
If specific affinities of retinal cell axons for particular cells in the corpora
quadrigemini were necessary for the reestablishment offunctionally opera-
tive contacts, the genetically determined neuron-to-neuron specificity
would necessarily be enormous. Recent evidence indicates that substantial
specificity of such reconnections does exist [12]. However, if, as in this
model, the output from the retinal cells is already coded when it leaves
the eye, then all that would be required is that some connection be es-
tablished, in which event the postulated decoder-recoder function in the
corpora quadrigemini would sort out patterns according to the site of
origin ofthe coded patterns rather than on the basis ofan overwhelmingly

608 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
complex series of(genetically determined) specific cell-to-cell connections.
Alternatively, of course, primary coding could be incorporated in the
corpora quadrigemini, but this would not rationalize Sperry's findings.
Seventh, it is interesting to speculate that the occurrence of color sen-
sations reportedly elicited by flashing a white light at suitable frequencies
on the human retina may depend on and reflect a similarity between the
patterns driven by the flashing frequency of discharge of retinal cells and
the sequences of pulses that represent the color coding signals normally
generated by color sensors and coders.
Eighth, the portion ofthe model dealing with image retention and with
attention versus alterations in brain wave pattern [7] appears to be con-
sistent with observation and with insight.
Ninth, the model would predict the existence of neurons of minute
size in the cortex so as to provide delay times corresponding to and de-
coding the time delays generated by anatomically larger coding cells and
sensory systems (a feature to be expected from the very fine axonic
projections which such cells possess) in intimate association with appro-
priate larger (and faster) distributor and collector neurons. The molecular
layer of the cortex, with its surrounding layers of large cells and lateral
and vertical tractules, seems to suit this prediction or deduction admirably.
The decoder elements may be the small cells (Ad, etc.), while the large
neurons and tractules may be the T-cell series as postulated in themodel
(fig. 15). In other words, the architecture of the cortex seems to be ad-
mirably suited to the requirements of this model [12].
Tenth, the model may account for a curious subjective phenomenon
I observed many years ago. During the initial stages of a tonsillectomy
anesthesia, the voice ofthe anesthetist evoked visual imagery (a descending
spiral of blue-white light). This image pulsated synchronously with the
vibrations in the auditory signal (the anesthetist's voice). Such an effect
might be expected if the auditory and visual cortex were differentially
anesthetized at that time and if the coded auditory signals were misinter-
preted because of their resemblance to the normal coding of visual sig-
nals (perhaps because of slowing or speeding up of transmission). Or,
alternatively, it might be that repressive inputs (via brain waves?) were
lifted, thus permitting a spillover of input into inappropriate decoders.
In this connection, the heightened association and perception ofvarious
sensations reportedly induced by hallucinogenic drugs—the visual-image-

609
evoking effects (spillover) of sounds, colors, music, etc.—might result
from the reduction in the number of elements required to fire the T2 or
T0 neurons. Selective drug effects (e.g., anesthesia, mood change, etc.)
might result from the selective tuning and detuning of central decoders
as the result of minute changes in conduction velocity that these sub-
stances induce in axons of different sizes which axonic branches may

Lamina zonalis

Lamina

pyramidalis ?
WJ
Lamina granulans »IP ^
interna"/'* ^
Lamina
ganglionaris

Lamina
W
multiformis

Fig. 15.—A section through the cerebral cortex, showing: A, The kinds ofconnections observed
between cells (by the Golgi method). B, The distribution of cells into discrete layers; large cells may
be T-type cells, and small cells (molecular or granular layer) correspond to the various decoder,
recoder, associational types (Ac, A4, Aa, etc.), (according to the Nissl method). C, The axonic path-
ways, suitable for the distributor-collector functions of the G?, Ti, etc., systems (Weigert method)
(after Herrick [13]).

??? Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969
impinge on a decoder system. The various disorders offunction in psychi-
atric patients may be due, in part, to an inability to suppress inappropriate
decoding or "ringing," due to brain wave inadequacy or to decreased
responsivity on the part ofrepressor sites which affect decoder thresholds.
Despite the foregoing facts and the interpretation of them afforded
by the model here described, it is evident that a series of critical tests of
the model's relevance should be carried out in the near future so that
useless projection and speculation may be avoided.
One feature of the model makes it particularly suitable for such test-
ing—the fact that the coders (and decoders-recoders) hypothesized must
retain their precise time-pattern relationship, not only with respect to the relative
times between elements in the coding sequence, but also in terms ofthe absolute
time sequences ofpulses which coders emit and to which decoders respond. There-
fore, any variable which produces an expansion, contraction, or distortion
of the time scale in either the coder or the decoder will result in the loss
of decoding ability or mis-decoding. One factor known to alter these
time relationships is the temperature of neurons, for the rate of axonic
conduction is substantially affected by temperature. Clearly, a primary
test of the relevance ofthe model to reality would be to determine whether
the experimental alteration of the relative temperature of suspected coders and
decoders will, infact, drastically interfere with message decoding—whether the
decoder has a higher or a lower temperature than the coding cell!
It may also be suggested that a central function of the circle of Willis,
is the maintenance of essentially equal temperatures in the various parts
of the brain in consequence of the thorough mixing of blood from both
of the carotid arteries this structure affords. It may also be suggested that
visual scotomas such as those that occur during migraine attacks are due
to the inability of cells to respond to signals appropriate to them because
of differential localized heating or cooling of the affected cortical areas
as a result of vascular spasms.
Further, it may be that the variable ability, from day to day, of senile
individuals [14] to recall (or otherwise function optimally, psychologi-
cally) is due to the occurrence of localized areas of heating, as a result of
the presence of local regions in which blood flow is somewhat impaired
because of cerebrovascular disease, and that the head-clearing effect of a
brisk walk is due to the achievement of a more equal temperature distri-

611
bution between various coder and decoder sites. Each of these concepts
should be capable of experimental resolution.
For the various reasons outlined above, the present model seems readily
susceptible to testing, for disproof or tentative confirmation, in the near
future. A paraphrase of Crick's concluding sentence in his paper on the
Wobble hypothesis might be appropriate here: "I should not be surprised
if the basic mechanisms suggested are not incorrect—and if the general
principle ofcoding-decoding, so central to molecular genetics, did have its
parallelisms at this higher level of pattern handling!" [15].
REFERENCES

i. J. D.Watson and F. H. C. Crick. Nature, 171:737, 1953.

2.R. Schweet et al. 19th Annu. M. D. Anderson Symp. Fundamental Cancer Research,
Austin, Texas. Austin: Univ. Texas Press, 1966.
3.H. R. Mahler and E.H. Cordes. Biological chemistry. New York: Harper & Row,
1966.
4.F.Jacob and J. Monad. J. MoI. Biol., 3:318, 1961.
5.B. L. Strehler, D. H. Hendley, and G. P. Hirsch. Nat. Acad. Sei. (U.S.), Proc,
S7:i75i, 1967.
6.K. S. Lashley. Physiol. Rev., 13:1, 1933.
7.H. W. MaGoun. The waking brain. 2d ed. Springfield, 111.: Charles C. Thomas,
1963.
8.Stuart Chase. The tyranny of words. New York: Harcourt, Brace, 1938.
9.K. D. Roeder. Sei. Amer., 212:94, 1965·
io. R. W. Sperry. In: M. Locke (ed.). The emergence of order in developing systems,
?. 306. New York: Academic, 1968.
11. ---------.J. Neurophysiol., 8:15, 1945.
12.Marcus Jacobson. In: M. Locke (ed.). Major problems in developmental biology,
p. 339. New York: Academic, 1966.
13.D.J. Cunningham. In: J. C. Brash and E. B. Jamieson (eds.). Textbook of anato-
my. 8th ed. New York: Oxford Univ. Press, 1943.
14.James E. Birren (ed.). Handbook of aging and the individual. Chicago: Univ.
Chicago Press, 1959.
15.F. H. C. Crick. J. MoI. Biol., 19:548, 1966.

612 Bernard L. Strehler · Information Handling in the Nervous System

Perspectives in Biology and Medicine · Summer 1969