In studying the history of ideas, it is interesting by klutzfu50


									  In studying the history of ideas, it is interesting to see how some concepts fallout of favour,
  then reappearconsiderablystrengthenedwith renewedvalidity from a fresh perspective.The insights
  of Gestalt psychology into visual perception are following exactly such a trajectory; after years
 of neglect, these insights are finding hard support in an unlikely place, in the details of the
  anatomy and physiology of the primary visual cortex (VI).
    In the early years of this century, concepts introduced by the school of Gestalt psychologists
  greatly influenced emergingideas aboutour perception complexentities-of music,of combinations
  of colours, and, particularly, of complex shapes, line drawings and figures. These conceptshinged
 around the idea of a "Gestalt", a perceptual"wholeness" of an object, that was somehow integral
 to its appearing as a single entity. This wholenesscould take many different forms (Wertheimer
  1938). The visual elements that made up a visual object or pattern could get bound together
 perceptually due to proximity. In figure la, for example, the pairs of neighbouring dots appear
 to group together and we automaticallysee the figure as a string of suchpairs. Perceptualgrouping
 could arise from similarity: in figure 1b, the dots of similar size automatically group with each
 other. The grouping could result from what Gestalt theorists termed "good continuation": in
 figure 1c we automatically see a sinusoid intersecting a squarewave. In all these examples,our
 perception of the particular groupings seems effortless. It takes some effort, in fact to see the
 elements grouped in some other way-to see figure la as a string of dots grouped into more
 distant pairs, or figure lc as a series of squarecorners and wavy arcs, or a string of boxes each
  with one wavy side.
    The problem with Gestalt explanationswas that it was hard to formulate Gestaltnessin analytical
 terms. You could tell when an object had good Gestalt but it was hard to think of increasing or
 decreasing the Gestalt or degree of wholeness in a manner open to experimental testing. This
 made the concept of Gestalt uncomfortably mystical. Moreover, with the technology of recording
 the responsesof single cortical neurons,starting in the '50's, physiologists beganto get a wealth
 of information at the other end of the scale of size-on how the brain dealt with the smallest
 elements of a percept. The study of single-cell responsesin the visual pathway, from the retina
 ,through the cortex, allowed one to see how complex sceneswere broken down to their simplest
 elements,to short line segments,edges,contrasts.The focus of researchshifted to understanding
 how the brain built up a percept hierarchically, starting with these minimal elements (Hubel and
 Wiesel 1962). For all of these reasons,Gestalt explanations largely fell out of favour.




                     Figure 1. Elements of an image group together by proximity (a),
                     similarity (b) or through "good continuation" (c).

                     J. Biosci., 24, No.   March 1999. @ lndian Ac~demy of Sciences
  2                                               Clipboard

     Starting with the late '80's, however, it became clear that the brain did not simply build up
  percepts in a "bottom-up" manner from the smallest visual elements. In many cases we appeared
  to first see an entire object in three-dimensional form before fully registering the smaller visual
  elements that made up the edges or details of the object (Nakayama and Shimojo 1990). One
  particular Gestalt attribute that got renewed attention and credibility was the concept of good
  continuation. The difference this time was that tests were designed using psychophysical methods
  to quantify and analyse the concept of good continuation. In a series of experiments, David Field
  and his associates used patterns made up of small elongated oblongs to quantify the attributes
   that made particular elements of visual scenes link up preferentially (Field et al 1993). In a typical
   experiment, they would first show subjects a target object, a "snake" made up of the oblong
  elements (figure 2a). Then, they would briefly show a visually cluttered image (figure 2b) which
  sometimes did and at other times did not contain the target snake. Subjects had to say whether
   or not they saw the snake amidst the clutter. Subjects got worse at picking out the snake from
  the clutter, in a smooth and measurable manner, as the separation, relative angle, lateral displacement
   or local curvature between the elements of the snake was increased (figure 2c, d). Field et al
   (1993) proposed that elements of a curve had an "association field" which linked the elements
   into a complete curve, perceptually, when they were close to each other, with relatively low
   curvature, no lateral displacement and no relative jitter between the angles made by the line
   elements (left side of figure 2e). This "association field" got quantifiably weaker as these conditions
  were progressively violated (right side of figure 2e).
     By this time, physiologists had also identified anatomical structures in primary visual cortex
  (VI) that seemed well suited to judge good continuation. In their initial work, Hubel and Wiesel
  (1962) showed that individual neurons in VI responded specifically to a well-defined receptive
  field (RF), that is, to a visual stimulus in a tightly defined area of visual space. Moreover, the
  stimulus had to' have specific properties-it     had to be a line element, oriented in a particular
  direction, of a particular colour, etc. These RF properties mapped smoothly over the surface of
  the cortex. Later, Gilbert and Wiesel (see Gilbert 1992 for a review) showed that neurons in VI
  sent out long axons that linked them to other neurons in VI, up to many mm away. These links
  were very specific: -neurons responding to line elements of. one particular orientation in one region
  of space were specifically linked to other neurons responding to line elements of the SAME
   orientation but in different regions of space. The effect of having such a network was as follows.
   Each neuron in VI still responded only when an optimally oriented line element fell within its
  RF. When different neurons in the large network had line elements in their individual RFs,
   however, they facilitated or inhibited each other's responses in a well-defined and predictable
   manner depending on the relative layout of the stimulus elements in space (Kapadia et al 1995).
      The response properties of such a network fit in very well with the requirements for judging
   good continuation. Through parallel studies in humans and monkeys Kapadia et al (1995) showed
   that the physiological responses of neurons in monkey V I to complex stimuli reflected precisely
   our perceptions of the same stimuli. The experimental question they asked was, how is our ability

                                                   ~          l         ...'.,
                                                                     '..-'..          I
                                                                    , "          .'
                                                                     , -.,', ,. -.,

          a                   b                                     d                     e
   Figure 2. (a, b) Easy "snake"; (c, d) difficult "snake.'; (e) pattern of the associationfield around a visual
   element, strong on the left and weak on the right.

                        J. Biosci.. 24. No.   March) 999. (!;) Indian Academy of Sciences

                                                 Clipboard                                                                    3

             iJ:                                                             \       \
                                                                                          1 1
                                                                                           1\                      j/   /
                                                                                                           ~;}, "
                                                                                     , f'
                                                                        \/                w-\
                                                                                                          '", \ w-\
                                                                         , ,               \ 1..-
                                                                                                            \ 'I

       a                      b                   c                          d                              e
Figure 3. (a) Single line element; (b) good flank; (c) poor flank; (d) element in noisy background; (e) added
collinear flanks.

to detect dim lines affected by putting other patterns in the background? In psychophysical
experiments they showed human subjects dim line elements in brief flashes (figure 3a); they varied
the brightness of the line elements and measured the brightness value (threshold) at which
subjects started just being able to detect the stimulus. When they showed subjects the dim target
line along with a bright flanking line (figure 3b), the threshold of detection of the dim target
line improved by 40%, i.e., the subjects were able to detect lines that were 40% dimmer than
the dimmest single line stimulus. This improvement in the detection threshold varied in a systematic
way with the position of the flanking line; a nearby collinear flank gave the strongest improvement
in the detection threshold. As the flank was moved further away, or to a different angle, or off
to a side, the detection threshold returned, progressively, to the threshold for the single line alone
(figure 3c). The experimenters then recorded from single neurons in monkey VI while the monkeys
were shown identical stimuli: either a single line element in the neuron's RF, or the element in
the RF along with another flanking element at various relative distances and positions. The response
of the neuron to a single line was enhanced up to 5-fold by .adding a nearby flanking line
(figure 3b). This facilitation was reduced as the flanking line was moved away, to a different
angle, or off to a side. The response to a single line was also suppressed by having a noisy
background (figure 3d); but the suppression could be overcome by adding collinear flanks (figure 3e).
These results make a strong case that facilitation through long-range horizontal connections in Vl
plays a significant role in our automatic perception of a "good" contour.
   So the apparently mystical and intrinsic Gestalt of good continuation that made particular contours
appear more whole could actually have a direct physiological base-in the anatomy and physiology
of long-range connections linking neurons in primary visual cortex.


Field D J, Hayes A and Hess R F 1993 Contour integration by the human visual system: evidence for a local
   "association field"; Vision Res. 33 173-193
Gilbert C D 1992 Horizontal integration and cortical dynamics; Neuron 9 1-13
Hubel D H and Wiesel T N 1962 Receptive fields, binocular interaction and functional architecture in the cat's
   visual cortex; J. Physiol. 160 106-154
Kapadia M K, Ito M, Gilbert C D and Westheimer G 1995 Improvement in visual sensitivity by changes in
   local context: parallel studies in human observers and in VI of alert monkeys; Neuron 15 843-856
Nakayama K and Shimojo S 1990 Toward a neural understanding of visual surface representation; Cold Spring
   Harbor Symp. Quant. Bioi. 55 911-924
Wertheimer M 1938 Laws of organization in perceptual forms (London: Harcourt)

                                                                                                          ANIRUDDHA         DAS
                                                                                                    Rockefeller University,
                                                                                                          /230 York Ave,
                                                                                           New      York, NY /0021, USA
                     J. Biosci., 24, No. I, March 1999. @ Indian Academy of Sciences
4                                              Clipboard

         Genomic imprinting:                   mom and dad (epi)genetics
 Genomic imprinting is an epigenetic chromosomal modification in the gametes or zygotes that
 results in a non-random monoallelic expression of specific autosomal genes depending upon their
 parent of origin. A gene could be maternally imprinted (only paternal copy expressed) or paternally
 imprinted (only maternal copy expressed). The exact mechanism of genomic imprinting is not
clear. However, allele-specific methylation of CpG residues at the 5' end of imprinted genes is
 thought to be responsible for the maintenance of imprinting in somatic cells.
    Several lines of evidence have pointed to the presence of imprinted genes in humans and other
mammals. Manipulated mouse embryos containing normal diploid chromosome numbers only from
the father or only from the mother fail to develop. Human concepti containing two normal sets
of genomes (2n = 46) only from the father fail to develop into embryos and instead produce
 hydatidiform moles. Human triploid abortuses are phenotypically different depending upon whether
the extra copy of the genome is from the father or the mother. The first genetic evidence of
genomic imprinting came from elegant breeding experiments in the mouse using chromosomal
rearrangements to produce uniparental disomy (UPD) for each chromosome and to assessphenotypic
effects on embryonic and early development (Cattanach and Kirk 1985). These experiments identified
imprinting effects for ten regions of six mouse chromosomes (see Ledbetter and Engel 1995).
 Deletion of certain specific chromosome regions in humans produces altogether different phenotypes
depending upon whether the deletion is on the maternal chromosome or the paternal chromosome.
 For example, deletion of the segment 15q11-13 on the paternal chromosome produces Prader-Willi
syndrome (small hands and feet, mild mental retardation, hypotonia, obesity, and hypogenitalism).
The same deletion on the maternal chromosome produces Angelman syndrome (severe mental
 retardation, seizures, hyperactivity and inappropriate laughter). Certain human genetic traits that
are autosomal dominant manifest only when inherited from one parent. For example, Beckwith-
 Wiedemann syndrome (gigantism, omphalocele, macroglossia and hemihypertrophy) is expressed
 only by people who inherit the mutation from their mother. It was not until 1991 that the presence
of imprinted genes was reported for the first time in the mouse genome through demonstration
of the selective maternal expression of Igf2r and H19, and the paternal expression of Igf2 (see
 Morison and Reeve 1998). Since then more than 20 imprinted genes have been identified in mice
and humans. In general, imprinted genes are found in a cluster, with the genes imprinted in
opposite directions. In the imprinted cluster on human chromosome 11p15.5, H19, KvLQT1,
IMPTl, HASH2, p57KIP2 and IPL are maternally expressed whereas IGF2 is paternally expressed.
Imprinted genes are known to show tissue-specific imprinting. KvLQT1 is imprinted in all tissues
examined so far except in the heart where it shows a biallelic expression. Therefore, the observation
of monoallelic expression of a gene in some tissues does not mean that the gene is imprinted in
all tissues. Moreover, a gene could be imprinted in one species and not in another. For example,
the mouse gene Igf2r is paternally imprinted whereas its human homologue IGF2R shows biallelic
expression. It is possible that the human gene IGF2R is imprinted in some tissues which have
 not yet been examined.
   Genomic imprinting is reversible and the imprinted status is reset during gametogenesis. For
example, in a male who receives an imprinted (inactive) allele of a gene from his father and an
expressed allele of the same gene from his mother, the imprinting will be reset for both his
alleles during spermatogenesis. In this case, when he passes either allele to his children, both
alleles will be imprinted. Similarly, in the case of a female who receives an imprinted allele of
the same gene from her father and an expressed allele from her mother, the imprinting process
will be reset during her oogenesis so that when she passes either allele to her children, they will
be expressed. Thus for a paternally imprinted gene, each individual will have one imprinted allele
coming from the father and one expressed allele from the mother. In a similar fashion, the
imprinted status of a maternally imprinted gene will be reset during gametogenesis of the mother
and the father so that all of their children will receive one imprinted allele from the mother and
one expressed allele from the father.

                      Biosc   24, No.1,   March 1999. @ Indian Academy of Science~

    Uniparental disomy (UPD), in which both members of a chromosome pair come from only one
parent, can unmask genomic imprinting and help to identify chromosomes or chromosome segments
with imprinted regions. An abnormal phenotype in cases with UPD could result from UPD per
se or from the homozygosity of a particular mutant allele present on both copies of chromosome
involved in UPD. Although it would be difficult to decide which one is true, an imprinting effect
is taken to be certain when three or more cases of UPD for a particular chromosome or chromosome
segment produce a similar constellation of abnormal phenotypes (Ledbetter and Engel 1995). It
should be noted that autosomal recessive effects are irrespective of imprinting. Based on known
UPD cases with abnormal phenotypes, Ledbetter and Engel (1995) have constructed a preliminary
imprinted map of the human genome. An imprinting effect is certain for four regions of the
human genome: maternal copy of chromosome 7q, paternal copy of chromosome IIp, maternal
 copy of chromosome 14q, and maternal and paternal copies of the proximal chromosome 15q
 (Ledbetter and Enge] ]995). Other possible regions of the human genome with imprinting effects
are maternal chromosome 2, paternal chromosome 6, paternal chromosome ]4, maternal chromosome
  16 and paterna] chromosome 20 (Ledbetter and Engel 1995). So far, imprinted genes from the
 proximal chromosome 15q (ZNF] 27, FNZ127, NDN, SNRPN, PAR-SN, PAR5, IPW, PAR 1,
 UBE3A, GABAA), chromosome IIp (H]9, IGF2, HASH2, KvLQT], p57KIP2, IPL, IMPTI, WT1),
chromosome 6q (MAS), chromosome 20 (GNAS), chromosome 7q (MEST), and chromosome 2
 (N-MYC) have been reported.
    The biological function of genomic imprinting is the subject' of considerable debate and remains
elusive. More than thirteen theories have been put forth to explain the biological and evolutionary
 significance of genomic imprinting (see Hurst and McVean 1997). According to the most popular
theory -the 'genome conflict' theory -genomic imprinting in placental mammals is the result of
 an evolutionary conflict between two sets of selfish genomes. Under conditions in which the
 paternity of future offspring from the same mother is not assured, the paternal genome would
 favour the removal of nutrients from the mother to the embryo (until such removal endangers the
 fitness of the embryo). On the other hand, irrespective of who the father is, all offspring born
 to a mother share their genes with her to the same extent. Therefore the maternal genome would
 tend to ~uppress such removal, preferring to keep more of the nutrief!cts in store for offspring yet
 to be born. Thus the conflict theory predicts that the expression of paternal genes in embryos
 tends to increase offspring size, whereas the expression of maternal genes tends to reduce it (see
 Hurst and McVean 1997). Paternal UPDll cases (Beckwith-Wiedemann syndrome patients) with
 large size, and maternal UPD7 and maternal UPD14 cases with growth retardation, support this
 theory. Paternal UPD6 cases with growth retardation on the other hand fail to support the theory.
    Genomic imprinting has received a great deal of attention in the past few years due to its
 role in several human genetic disorders such as Prader-Willi syndrome, Angelman syndrome,
  Beckwith-Wiedemann syndrome, gestational trophoblastic diseases and cancer. An imprinted gene,
 IPL, has recently been isolated from human chromosome 11p15.5 with similarity to TDAG51
  which has been implicated in Fas expression and apoptosis (Qian et at ]997). This suggests that
 genomic imprinting may also have a role in apoptosis. Perhaps the most striking observation
concerning the consequencesof imprinting has recently been made by Lefebvre et at (1998)
regarding the role of the imprinted gene Mest (Peg I) in mice. This gene is expressed only from
the paternal allele. Lefebvre et at (1998) introduced a mutation in Mest by gene targeting in
embryonic stem (ES) cells and obtained Mest-deficient mice. The loss of Mest function not only
resulted in intrauterine growth retardation (IUGR), increased prenatal and postnatal lethality and
decreased reproductive fitness of females, but also abnormal maternal behaviour. Normal maternal
behaviour in mice includes responding appropriately to newborns, ingesting their extraembryonic
tissues after parturition (placentophagia), and influences the time required to retrieve one of the
three pups as well as the latency to initiate the construction of a nest with wood chips. Most or
all pups born to Mest-deficient female mice were left untouched after parturition, with almost
complete absence of placentophagia. Where wild-type females displayed an immediate inclination
for nest building, the Mest-deficient female mice performed poorly. Wild-type female mice retrieved
the first pup in -] 0 min whereas Mest-deficient female mice did not show this behaviour within
the 15 min period of observation (Lefebvre et at 1998).

                      Biosc   24, No.   March 1999. @ Indian Academy of Sciences
6                                               Clipboard
  The human homologue of the mouse Mest gene is located on chromosome 7q32. MEST is
paternally expressed in human fetal tissues, but biallelically expressed in adult lymphocytes.
Maternal UPD7 cases in humans demonstrate intrauterine and postnatal growth retardation. This
suggests that the phenotypic consequences of the loss of MEST function are similar in humans
and mice with regard to intrauterine and postnatal growth retardation. Thus Lefebvre et al (1998),
with their elegant experiment in mice using the paternally expressed gene Mest, have shown for
the first time the role of genomic imprinting in animal behaviour. As more imprinted genes are
isolated from humans and other mammals, and their roles elucidated, one hopes that the biological
purpose of genomic imprinting will not remain elusive for long.

Cattanach B M and Kirk M 1985 Differential activity of maternally and paternally derived chromosome regions
  in mice; Nature (London) 315 496--498
Hurst L D and McVean G T 1997 Growth effects of uniparental disomies and the conflict theory of genomic
  imprinting; Trends Genet. 13 436--443
Ledbetter D H and Engel E 1995 Uniparental disomy in humans: development of an imprinted map and its
  implications for prenatal diagnosis; Hum. Mol. Genet. 4 1757-1764
Lefebvre L, Viville S, Barton S C, Ishino F, Keverne E Band Surani M A 1998 Abnormal maternal behavior
  and growth retardation associated with loss of the imprinted gene Mest; Nature Genet. 20 163-169
Morison I M and Reeve A E 1998 A catalogue of imprinted genes and parent-of-origin effects in humans and
  animals; Hum. Mol. Genet. 7. 1599-1609
Qian N, Frank D, O'Keefe D, Dao D, ZJ1ao L, Yuan L, Keating M, Walsh C and Tycko B 1997 The IPL
  gene on chromosome Ilp15.5 is imprinted in humans and mice and is similar to TDAG51, implicated in
    Fas expression and apoptosis: Hum. Mol. Genet. 6 2021-2029

                                                                                             ARUN KUMAR
                                                                  Department of Molecular Reproduction,
                                                                               Development and Genetic~',
                                                In{iian IIJ.\"titute of Science, Bangalore 560 012, I1Jdia

                         Biosci.. 24, No. i, March 1999. @ indian Academy of Sciences

To top