General Explanations A run counts as a false negative in the superset sense if neither the injected assembly nor a superset of it exists in the filtered/reduced pattern set. A run counts as a false negative in the exact sense if the injected assembly does not exist exactly in the filtered/reduced pattern set. The expression “all other patterns” refers to any sets of neurons that are not identical to the injected assembly (that is, proper supersets and proper subsets of the injected assembly, sets overlapping with the injected assembly, and sets unrelated to the injected assembly). The bar charts show (1) the decimal logarithm of the average number of patterns found in surrogate data (independent Poisson processes); (2) the fraction of runs in which an injected assembly was not detected (neither exactly nor as a superset, false negatives in the superset sense); (3) the fraction of runs in which an injected assembly was not detected (false negatives in the exact sense); and (4) the fraction of runs in which any other pattern was reported, classified according to the size and the number of coincidences of the injected assembly. For a more detailed analysis, (non-exact) detections (that is, any patterns other than the injected assembly itself) are categorized into four classes: (1) superset patterns: the pattern is a proper superset of the injected assembly (that is, all neurons of the injected assembly are present and there is at least one additional neuron, a so-called “excess neuron”); (2) subset patterns: the pattern is a proper subset of the injected assembly (that is, at least one neuron of the assembly is missing); (3) overlap patterns: the pattern contains at least two, but not all neurons from the injected assembly and at least one other neuron; (4) unrelated patterns: patterns that have none or at most one neuron in common with the injected assembly. The bar charts corresponding to these pattern categories show the decimal logarithm of the average number of patterns found in a run, classified according to the size and the number of coincidences of the injected assembly. The reason for allowing unrelated patterns to share one neuron with the injected assembly is that the assembly activity only increases the chance of coincident spiking events of patterns that share at least two neurons with the assembly, because only then the coincidences of the assembly can have an influence on the chance of coincidences of the overlapping pattern. It should be noted that only the additional patterns are true false positives. All other pattern types are induced by the injected assembly and occur due to (1) one or more neurons outside the injected assembly accidentally firing together with a few of the coincident spiking events of the injected assembly (superset patterns); (2) neurons in a subset of the injected assembly firing together one or more times in addition to the coincident spiking events of the injected assembly due to background spiking (subset patterns); (3) like 2, but with one or more neurons outside of the injected assembly firing together with a few of the coincident spiking events of the injected assembly and at least one of the additional spiking events of the subset created by background spiking (overlap patterns). Parameters: 20Hz firing rate for all neurons (corrected for injected coincident spikes). 3s recording period, discretized with time windows of 2, 3, 4, or 5ms. 10,000 surrogate data sets for the frequent pattern bar chart. 1000 runs for each bar of the other bar charts. 1
37
Embed
General Explanations - borgelt.net · General Explanations A run counts as a false negative in the superset sense if neither the injected assembly nor a superset of it exists in the
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
General Explanations
A run counts as a false negative in the superset sense if neither the injected assembly nor asuperset of it exists in the filtered/reduced pattern set. A run counts as a false negative in theexact sense if the injected assembly does not exist exactly in the filtered/reduced pattern set.The expression “all other patterns” refers to any sets of neurons that are not identical to theinjected assembly (that is, proper supersets and proper subsets of the injected assembly, setsoverlapping with the injected assembly, and sets unrelated to the injected assembly).
The bar charts show (1) the decimal logarithm of the average number of patterns found insurrogate data (independent Poisson processes); (2) the fraction of runs in which an injectedassembly was not detected (neither exactly nor as a superset, false negatives in the supersetsense); (3) the fraction of runs in which an injected assembly was not detected (false negativesin the exact sense); and (4) the fraction of runs in which any other pattern was reported,classified according to the size and the number of coincidences of the injected assembly.
For a more detailed analysis, (non-exact) detections (that is, any patterns other than theinjected assembly itself) are categorized into four classes: (1) superset patterns: the pattern is aproper superset of the injected assembly (that is, all neurons of the injected assembly are presentand there is at least one additional neuron, a so-called “excess neuron”); (2) subset patterns: thepattern is a proper subset of the injected assembly (that is, at least one neuron of the assemblyis missing); (3) overlap patterns: the pattern contains at least two, but not all neurons from theinjected assembly and at least one other neuron; (4) unrelated patterns: patterns that have noneor at most one neuron in common with the injected assembly. The bar charts corresponding tothese pattern categories show the decimal logarithm of the average number of patterns found ina run, classified according to the size and the number of coincidences of the injected assembly.The reason for allowing unrelated patterns to share one neuron with the injected assembly isthat the assembly activity only increases the chance of coincident spiking events of patterns thatshare at least two neurons with the assembly, because only then the coincidences of the assemblycan have an influence on the chance of coincidences of the overlapping pattern.
It should be noted that only the additional patterns are true false positives. All other patterntypes are induced by the injected assembly and occur due to (1) one or more neurons outsidethe injected assembly accidentally firing together with a few of the coincident spiking events ofthe injected assembly (superset patterns); (2) neurons in a subset of the injected assembly firingtogether one or more times in addition to the coincident spiking events of the injected assemblydue to background spiking (subset patterns); (3) like 2, but with one or more neurons outside ofthe injected assembly firing together with a few of the coincident spiking events of the injectedassembly and at least one of the additional spiking events of the subset created by backgroundspiking (overlap patterns).
Parameters:20Hz firing rate for all neurons (corrected for injected coincident spikes).3s recording period, discretized with time windows of 2, 3, 4, or 5ms.10,000 surrogate data sets for the frequent pattern bar chart.1000 runs for each bar of the other bar charts.
1
No Pattern Set Reduction
Let S be the set of signatures that occur in the surrogates. All patterns that are left over afterprimary pattern filtering are kept, that is, all patterns remaining after removing all sets I withsignatures 〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S.
2ms jitter, 2ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
122
46
810
12
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
2
3ms jitter, 3ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
3
Pattern Set Reduction with Excess Coincidences 1
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff
• 〈zB, cB − cA〉 = 〈|B|, s(B)− s(A)〉 ∈ S.
Otherwise B is preferred to A. In other words: A is preferred to B only if the excess coincidencesof B can be explained (heuristically) as chance events.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess coincidences 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4
3ms jitter, 3ms windows, reduced with excess coincidences 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess coincidences 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess coincidences 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5
Pattern Set Reduction with Excess Coincidences 2
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff
• 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 ∈ S.
Otherwise B is preferred to A. In other words: A is preferred to B only if the excess coincidencesof B can be explained (heuristically) as chance events.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess coincidences 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
6
3ms jitter, 3ms windows, reduced with excess coincidences 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess coincidences 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess coincidences 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
7
Pattern Set Reduction with Excess Neurons
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set B is preferred to the set A iff
• 〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 ∈ S.
Otherwise A is preferred to B. In other words: B is preferred to A only if the excess neurons ofA can be explained (heuristically) as chance events.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess neurons
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
8
3ms jitter, 3ms windows, reduced with excess neurons
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess neurons
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess neurons
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
9
Reduction with Number of Covered Spikes 1
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff
• zAcA ≥ zBcB.
Otherwise B is preferred to A. In other words: A is preferred to B only if it covers at leastas many spikes as B (assuming that more covered spikes make a pattern less likely to occur bychance).
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
122
46
810
12
unrelatedpatterns
10
3ms jitter, 3ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
11
Reduction with Number of Covered Spikes 2
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff
• (zA − 1)cA ≥ (zB − 1)cB.
Otherwise B is preferred to A. In other words: A is preferred to B only if it covers at least asmany “coincident” spikes as B (assuming that spikes, in order to be coincident, need a referenceto be coincident to, which itself should not be counted).
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
122
46
810
12
unrelatedpatterns
12
3ms jitter, 3ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
13
Reduction with Excess Coincidences and Excess Neurons 1
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 ∈ S and
• 〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 /∈ S or zAcA ≥ zBcB.
The set B is preferred to the set A iff 〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 ∈ S and
• 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 /∈ S or zAcA < zBcB.
Otherwise A and B are not comparable (that is, neither is preferred to the other). In otherwords: A is preferred to B if the excess coincidences of B can be explained (heuristically) aschance events, but the excess neurons in A cannot be explained (heuristically) as chance events.Analogously, B is preferred to A if the excess neurons of A can be explained (heuristically) aschance events, but the excess coincidences of B cannot be explained (heuristically) as chanceevents. If both the excess neurons of A and the excess coincidences of B can be explainedas chance events, the number of covered spikes is invoked as a secondary criterion to make adecision. Finally, if neither the excess neurons of A nor the excess coincidences of B can beexplained (heuristically) as chance events, no preference relation is established.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
14
3ms jitter, 3ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
15
Reduction with Excess Coincidences and Excess Neurons 2
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 ∈ S and
• 〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 /∈ S or (zA − 1)cA ≥ (zB − 1)cB.
The set B is preferred to the set A iff 〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 ∈ S and
• 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 /∈ S or (zA − 1)cA < (zB − 1)cB.
Otherwise A and B are not comparable (that is, neither is preferred to the other). In otherwords: A is preferred to B if the excess coincidences of B can be explained (heuristically) aschance events, but the excess neurons in A cannot be explained (heuristically) as chance events.Analogously, B is preferred to A if the excess neurons of A can be explained (heuristically) aschance events, but the excess coincidences of B cannot be explained (heuristically) as chanceevents. If both the excess neurons of A and the excess coincidences of B can be explained aschance events, the number of covered coincident spikes is invoked as a secondary criterion tomake a decision. Finally, if neither the excess neurons of A nor the excess coincidences of B canbe explained (heuristically) as chance events, no preference relation is established.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
16
3ms jitter, 3ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
17
Reduction with Excess Coincidences and Excess Neurons 3
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff
• 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 ∈ S and〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 /∈ S or
• (〈zB, cB − cA + 1〉 ∈ S = 〈zA − zB + 2, cA〉 ∈ S) and zAcA ≥ zBcB.
Otherwise B is preferred to A. In other words: A is preferred to B if both the excess coincidencespreference relation and the excess neurons preference relation prefer A to B; and B is preferredto A if both the excess coincidences preference relation and the excess neurons preference relationprefer B to A. Finally, if the two preference relations disagree, the number of covered spikesestablishes the preference. Or, more concisely: if the excess coincidences and excess neuronspreference relations agree, they define the preference. If they disagree, the number of coveredspikes is invoked to make a decision.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess coincidences and neurons 3
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
18
3ms jitter, 3ms windows, reduced with excess coincidences and neurons 3
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess coincidences and neurons 3
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess coincidences and neurons 3
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
19
Reduction with Excess Coincidences and Excess Neurons 4
Let S be the set of signatures that occur in the surrogates. Let A and B with B ⊂ A be twosets left over after primary pattern filtering, that is, after removing all sets I with signatures〈zI , cI〉 = 〈|I|, s(I)〉 ∈ S, and therefore 〈zA, cA〉 /∈ S and 〈zB, cB〉 /∈ S.
The set A is preferred to the set B iff
• 〈zB, cB − cA + 1〉 = 〈|B|, s(B)− s(A) + 1〉 ∈ S and〈zA − zB + 2, cA〉 = 〈|A| − |B|+ 2, s(A)〉 /∈ S or
• (〈zB, cB − cA + 1〉 ∈ S = 〈zA − zB + 2, cA〉 ∈ S) and (zA − 1)cA ≥ (zB − 1)cB.
Otherwise B is preferred to A. In other words: A is preferred to B if both the excess coincidencespreference relation and the excess neurons preference relation prefer A to B; and B is preferredto A if both the excess coincidences preference relation and the excess neurons preference relationprefer B to A. Finally, if the two preference relations disagree, the number of covered coincidentspikes establishes the preference. Or, more concisely: if the excess coincidences and excessneurons preference relations agree, they define the preference. If they disagree, the number ofcovered coincident spikes is invoked to make a decision.
Pattern set reduction keeps only sets for which there exists no subset and no superset that ispreferred to them.
2ms jitter, 2ms windows, reduced with excess coincidences and neurons 4
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
20
3ms jitter, 3ms windows, reduced with excess coincidences and neurons 4
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
4ms jitter, 4ms windows, reduced with excess coincidences and neurons 4
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
122
46
810
12
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
5ms jitter, 5ms windows, reduced with excess coincidences and neurons 4
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
all otherpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
supersetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
subsetpatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
overlappatterns
coincidencesc
asse
mblysiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
24
68
10
12
24
68
1012
unrelatedpatterns
21
3ms jitter, 3ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
3ms jitter, 3ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
22
4ms jitter, 4ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
4ms jitter, 4ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
23
5ms jitter, 5ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
5ms jitter, 5ms windows, reduced with number of covered spikes 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
24
3ms jitter, 3ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
3ms jitter, 3ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
25
4ms jitter, 4ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
4ms jitter, 4ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
26
5ms jitter, 5ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
5ms jitter, 5ms windows, reduced with number of covered spikes 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
27
3ms jitter, 3ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
3ms jitter, 3ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
28
4ms jitter, 4ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
4ms jitter, 4ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
29
5ms jitter, 5ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
5ms jitter, 5ms windows, reduced with excess coincidences and neurons 1
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
30
3ms jitter, 3ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
3ms jitter, 3ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
31
4ms jitter, 4ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
4ms jitter, 4ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
32
5ms jitter, 5ms windows, filtered with surrogate data
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
5ms jitter, 5ms windows, reduced with excess coincidences and neurons 2
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.super
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons8 coins.
coincidencesc
asse
mblysiz
e z
rate
0
0.2
0.4
0.6
0.8
1
24
68
10
12
24
68
1012
false neg.exact
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
33
Average pattern counts, computed from 1000 runs
3ms jitter, 3ms windows, reduced with number of covered spikes 2
surrogates unfiltered filtered reduced
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
34
Average pattern counts, computed from 1000 runs
4ms jitter, 4ms windows, reduced with number of covered spikes 2
surrogates unfiltered filtered reduced
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
35
Average pattern counts, computed from 1000 runs
5ms jitter, 5ms windows, reduced with number of covered spikes 2
surrogates unfiltered filtered reduced
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
5 neurons5 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
6 neurons6 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
7 neurons7 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
122
46
810
12
8 neurons8 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–4
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
patternspectrum
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
coincidencesc
patte
rnsiz
e z
log(
#pat
tern
s)
–3
–2
–1
0
1
2
3
24
68
10
12
24
68
1012
9 neurons9 coins.
36
Pattern counts for a single run
3ms jitter, 3ms windows, reduced with number of covered spikes 2