Fourteen adult cats were used for this study. Anesthesia was initiated
with ketamine (15mg/kg) and xylazine (1.5 mg/kg) i.m., and maintained with
isofluorane (typically 0.6-1.5% in 70/30 N
O/O
) delivered through a
tracheal cannula. Cats were paralyzed with a cocktail of gallamine
triethiodide (3.6 mg/hour) and tubocurarine chloride (0.15 mg/hour), and
artificially respired to maintain end-tidal CO
at a partial pressure of
mm Hg. EEG and heart rate were continuously monitored to ensure
adequate anesthesia. After a craniotomy to expose visual cortex, the
recording chamber was centered at Horsley-Clark coordinate P2, filled
with silicone oil and sealed with a quartz plate. The location of area 17
was confirmed by histology in previous experiments
[33].
Techniques for intrinsic
signal imaging were briefly as follows. Images were obtained from a
depth of 
m under 610nm light, using a tandem-lens macroscope
arrangement giving an overall magnification of 75 pixels/mm. Images were
collected using a Bischke CCD-5024N video camera at a 655x480 resolution
without binning, and digitized by an Imager 2001 system (Optical Imaging Inc.)
and a 486-66 PC with a Matrox graphics board. Frames were summed
between 1.3 and 4.0 sec after onset of stimulus motion, corresponding to
the time of maximum signal as determined by our previous experiments. Data
was analyzed using programs written by the authors (L.T., C.R.) using C++
(Borland) or IDL (Research Systems Inc. Boulder, CO).
Gratings (spatial frequency 1.0 cyc/
, drifting at 1.5 
/sec)
were shown
on a 17 inch monitor positioned 28.5 cm in front of the animal. Grating
contrast was 99%; neutral grey intensity was 6.0 cd/
.
Single-units were
recorded at the conclusion of some imaging experiments using
electrodes of 
impedance (Micro Probe Inc., Hagerstown, MD).
Receptive fields were plotted by collecting spike responses to short
moving bars and measuring the minimum response
field. All stimuli were shown to the contralateral eye only. Eye position
was checked regularly during the course of the experiment by using a
reversing ophthalmoscope to project an image of the retinal vasculature
onto the monitor.
Gratings of 4 or 8 orientations per stimulus type (``center'', ``surround'' or ``full-field'') were shown in an interleaved fashion 40-80 times over a period of 2-6 hours. The size of the ``center'' stimulus (and the complementary ``surround'') was systematically varied in early experiments and adjusted to cover the aggregate classical receptive field of neurons within the center representation (see figure 6). Maps of orientation preference were obtained by dividing two maps of orthogonal stimulus orientations. To obtain composite maps, the vector average of the signal was computed pixel by pixel, and the resulting map displayed in ``polar'' form using color to code the vector angle and the intensity of each color to code the vector magnitude (see [6]).