\relax 
\citation{Magee2000}
\citation{Iansek1973a}
\citation{Magee2000}
\citation{Rapp1994}
\citation{Rapp1994}
\citation{Hines1997}
\citation{Rapp1994}
\citation{Rapp1994}
\citation{Hines1997}
\citation{Rapp1992}
\citation{BorgGraham1998}
\citation{Pare1998}
\citation{Kamondi1998}
\citation{BorgGraham1998}
\citation{Pare1998}
\citation{Kamondi1998}
\@writefile{lof}{\contentsline {figure}{\numberline {1}{\ignorespaces \baselineskip 11pt\relax  { \relax \fontsize  {10}{12}\selectfont  \abovedisplayskip 10\p@ plus2\p@ minus5\p@ \abovedisplayshortskip \z@ plus3\p@ \belowdisplayshortskip 6\p@ plus3\p@ minus3\p@ \def \leftmargin \leftmargini \parsep 4.5\p@ plus2\p@ minus\p@ \topsep 9\p@ plus3\p@ minus5\p@ \itemsep 4.5\p@ plus2\p@ minus\p@ {\leftmargin \leftmargini \topsep 6\p@ plus2\p@ minus2\p@ \parsep 3\p@ plus2\p@ minus\p@ \itemsep \parsep }\belowdisplayskip \abovedisplayskip \fontfamily  {cmss}\baselineskip 11pt\relax \selectfont  {\bf  \fontfamily  {cmss}\baselineskip 11pt\relax \selectfont  Network activity eliminates the location-independence of somatic EPSP amplitude found in vitro.} {\bf  (a)} Left. Model of two-months old CA1 pyramidal neuron (courtesy of D.Turner)with two synaptic inputs located $150\tmspace  +\thinmuskip {.1667em}\mu m$ (proximal) and $450\tmspace  +\thinmuskip {.1667em}\mu m$ (distal) from the soma (scaling bar, $100\tmspace  +\thinmuskip {.1667em}\mu m$). The area of $15,000$ dendritic spines (1 $\mu m^2$ each) was incorporated globally into the model\nobreakspace  {}\citep  {Rapp1994}. Top right. Simulating the {\em  in-vitro} case, the proximal and distal inputs give rise to location-independent somatic EPSP amplitude ($0.2\tmspace  +\thinmuskip {.1667em} mV$) due to 4.1 fold increase in the amplitude of the synaptic conductance ($g_{syn}$) for the distal input (reference model parameters are $R_m =40,000\tmspace  +\thinmuskip {.1667em} \Omega cm^2$ ,$R_i =150\tmspace  +\thinmuskip {.1667em} \Omega cm$ , $C_m =1\tmspace  +\thinmuskip {.1667em}\mu F/{cm}^2$). Bottom right. In the simulated {\em  in-vivo} case, 15,680 excitatory synaptic inputs contact the model neuron, each is activated randomly once per second. As a result of this 'background network activity', the amplitude of the somatic EPSP from the distal synapse becomes smaller by a factor of 2 compared to that from the proximal synapse. {\bf  (b)} Profile of $g_{syn}(x)$ that is required to produce location-independent EPSPs of $0.2\tmspace  +\thinmuskip {.1667em} mV$ at the soma for all dendritic compartments up to $650\tmspace  +\thinmuskip {.1667em}\mu m$ from the soma. Results for two different combinations of model parameters are shown. (Curves are a result of exponential fit). {\bf  (c)} Somatic EPSP amplitude as a function of distance of the synapse from the soma, for the reference 'location-independent' '{\em  in-vitro}' case (horizontal dashed line) and for several simulated '{\em  in-vivo}' cases. Black line: the input resistance of the reference in vitro case was reduced 4-fold by increasing $G_m$ uniformly (6-fold). The other 3 curves show the effect of the mutual synaptic shunt expected {\em  in vivo}, taking into account the {\em  in vitro} profile of $g_{syn}(x)$ for a variety of model parameters. For one case (red), somatic EPSPs amplitude from all dendritic compartments are shown (red dots) with the corresponding quadratic fit (red line). The dramatic decrease in the somatic EPSP amplitude from distal inputs is clearly seen in all cases. The impact of network activity on the dendritic membrane conductance was modeled by incorporating into the resting $G_m$, at each dendritic compartment, the corresponding average gsyn that gives rise to location-independent EPSP in the '{\em  in-vitro}' condition\nobreakspace  {}\citep  {Rapp1994}. AMPA-like unitary synaptic input was simulated as transient conductance change with $1\tmspace  +\thinmuskip {.1667em}ms$ time-to-peak and a driving force of $65\tmspace  +\thinmuskip {.1667em}mV$. Specific membrane resistance, $R_m$ (in $\Omega {cm}^2$), axial resistance, $R_i$ (in $\Omega cm$) and input rate (in $Hz$) are denoted in the corresponding curves. Simulations were performed using NEURON\nobreakspace  {}\citep  {Hines1997}. }}}{84}}
\newlabel{fig:LocIndepFig1}{{1}{83}}
\citation{Reyes2001}
\citation{Kamondi1998}
\citation{Graham1999}
\citation{Poirazi2001}
\@writefile{lof}{\contentsline {figure}{\numberline {2}{\ignorespaces \baselineskip 11pt\relax  \relax \fontsize  {10}{12}\selectfont  \abovedisplayskip 10\p@ plus2\p@ minus5\p@ \abovedisplayshortskip \z@ plus3\p@ \belowdisplayshortskip 6\p@ plus3\p@ minus3\p@ \def \leftmargin \leftmargini \parsep 4.5\p@ plus2\p@ minus\p@ \topsep 9\p@ plus3\p@ minus5\p@ \itemsep 4.5\p@ plus2\p@ minus\p@ {\leftmargin \leftmargini \topsep 6\p@ plus2\p@ minus2\p@ \parsep 3\p@ plus2\p@ minus\p@ \itemsep \parsep }\belowdisplayskip \abovedisplayskip \fontfamily  {cmss}\baselineskip 11pt\relax \selectfont  {\bf  \fontfamily  {cmss}\baselineskip 11pt\relax \selectfont  Attempting to restore location-independence of somatic EPSP amplitude {\em  in-vivo} using the mechanism of synaptic scaling.} {\bf  (a,b)} Temporally coherent and spatially stratified synaptic activation. Model tree was divided into 10 iso-distant layers ($85\tmspace  +\thinmuskip {.1667em}\mu m$ each). Within each layer,synapses were divided into groups of a given size and each group was activated synchronously by the same random ($1\tmspace  +\thinmuskip {.1667em} Hz$) spike train, and independently from all other groups. {\bf  (b)} The normalized somatic EPSP amplitude for three different group sizes and for the reference case (black line) in which all dendritic synapses where activated randomly and asynchronously at $1\tmspace  +\thinmuskip {.1667em} Hz$. {\bf  (c)} Synaptic scaling (continuous red line) that preserves in vivo location independence for a target $0.2\tmspace  +\thinmuskip {.1667em} mV$ soma EPSP (horizontal red dots) with asynchronous background frequency of $1\tmspace  +\thinmuskip {.1667em}Hz$; the corresponding synaptic scaling for the '{\em  in vitro}' case is shown by the blue line. Increasing '{\em  in-vivo}' background frequency to $3\tmspace  +\thinmuskip {.1667em} Hz$ destroys location independence (green dots). Algorithm for obtaining location independence {\em  in-vivo} starts with (i) the synaptic conductance that maintains location independence {\em  in-vitro}. (ii)For a given asynchronous {\em  in-vivo} condition, the average synaptic conductance for each compartment is incorporated to the resting membrane conductance. (iii) The somatic EPSP amplitude for each synapse is measured individually and the corresponding synaptic conductance is scaled in proportion to the deviation from the target value ($0.2\tmspace  +\thinmuskip {.1667em}mV$). Iterations through steps (ii) and (iii) continues until convergence is obtained. }}{86}}
\newlabel{fig:LocIndepFig2}{{2}{85}}
\bibstyle{apalike}
\bibdata{Refs}
\bibcite{BorgGraham1998}{{1}{1998}{{Borg-Graham et~al.}}{{}}}
\bibcite{Graham1999}{{2}{1999}{{Graham}}{{}}}
\bibcite{Hines1997}{{3}{1997}{{Hines and Carnevale}}{{}}}
\bibcite{Iansek1973a}{{4}{1973}{{Iansek and Redman}}{{}}}
\bibcite{Kamondi1998}{{5}{1998}{{Kamondi et~al.}}{{}}}
\bibcite{Magee2000}{{6}{2000}{{Magee and Cook}}{{}}}
\bibcite{Pare1998}{{7}{1998}{{Pare et~al.}}{{}}}
\bibcite{Poirazi2001}{{8}{2001}{{Poirazi and Mel}}{{}}}
\bibcite{Rapp1994}{{9}{1994}{{Rapp et~al.}}{{}}}
\bibcite{Rapp1992}{{10}{1992}{{Rapp et~al.}}{{}}}
\bibcite{Reyes2001}{{11}{2001}{{Reyes}}{{}}}