diff --git a/beacon_chain/consensus_object_pools/attestation_pool.nim b/beacon_chain/consensus_object_pools/attestation_pool.nim
index 6a995da4d..711ae9279 100644
--- a/beacon_chain/consensus_object_pools/attestation_pool.nim
+++ b/beacon_chain/consensus_object_pools/attestation_pool.nim
@@ -280,6 +280,12 @@ func updateAggregates(entry: var AttestationEntry) =
             inc j
         inc i
 
+func covers(entry: AttestationEntry, bits: CommitteeValidatorsBits): bool =
+  for i in 0..<entry.aggregates.len():
+    if bits.isSubsetOf(entry.aggregates[i].aggregation_bits):
+      return true
+  false
+
 proc addAttestation(entry: var AttestationEntry,
                     attestation: Attestation,
                     signature: CookedSig): bool =
@@ -304,12 +310,8 @@ proc addAttestation(entry: var AttestationEntry,
     entry.singles[singleIndex.get()] = signature
   else:
     # More than one vote in this attestation
-    for i in 0..<entry.aggregates.len():
-      if attestation.aggregation_bits.isSubsetOf(entry.aggregates[i].aggregation_bits):
-        trace "Aggregate already seen",
-          singles = entry.singles.len(),
-          aggregates = entry.aggregates.len()
-        return false
+    if entry.covers(attestation.aggregation_bits):
+      return false
 
     # Since we're adding a new aggregate, we can now remove existing
     # aggregates that don't add any new votes
@@ -376,6 +378,24 @@ proc addAttestation*(pool: var AttestationPool,
   if not(isNil(pool.onAttestationAdded)):
     pool.onAttestationAdded(attestation)
 
+func covers*(
+    pool: var AttestationPool, data: Attestationdata,
+    bits: CommitteeValidatorsBits): bool =
+  ## Return true iff the given attestation already is fully covered by one of
+  ## the existing aggregates, making it redundant
+  ## the `var` attestation pool is needed to use `withValue`, else Table becomes
+  ## unusably inefficient
+  let candidateIdx = pool.candidateIdx(data.slot)
+  if candidateIdx.isNone:
+    return false
+
+  let attestation_data_root = hash_tree_root(data)
+  pool.candidates[candidateIdx.get()].withValue(attestation_data_root, entry):
+    if entry[].covers(bits):
+      return true
+
+  false
+
 proc addForkChoice*(pool: var AttestationPool,
                     epochRef: EpochRef,
                     blckRef: BlockRef,
diff --git a/beacon_chain/gossip_processing/gossip_validation.nim b/beacon_chain/gossip_processing/gossip_validation.nim
index 4f2e63ec4..d8bd298d2 100644
--- a/beacon_chain/gossip_processing/gossip_validation.nim
+++ b/beacon_chain/gossip_processing/gossip_validation.nim
@@ -621,9 +621,24 @@ proc validateAggregate*(
       return err(v.error)
     v.get()
 
-  # [REJECT] aggregate_and_proof.selection_proof selects the validator as an
-  # aggregator for the slot -- i.e. is_aggregator(state, aggregate.data.slot,
-  # aggregate.data.index, aggregate_and_proof.selection_proof) returns True.
+  if pool[].covers(aggregate.data, aggregate.aggregation_bits):
+    # https://github.com/ethereum/consensus-specs/issues/2183 - althoughh this
+    # check was temporarily removed from the spec, the intent is to reinstate it
+    # per discussion in the ticket.
+    #
+    # [IGNORE] The valid aggregate attestation defined by
+    # `hash_tree_root(aggregate)` has _not_ already been seen
+    # (via aggregate gossip, within a verified block, or through the creation of
+    # an equivalent aggregate locally).
+
+    # Our implementation of this check is slightly different in that it doesn't
+    # consider aggregates from verified blocks - this would take a rather heavy
+    # index to work correcly under fork conditions - we also check for coverage
+    # of attestation bits instead of comparing with full root of aggreagte:
+    # this captures the spirit of the checkk by ignoring aggregates that are
+    # strict subsets of other, already-seen aggregates.
+    return errIgnore("Aggregate already covered")
+
   let
     epochRef = block:
       let tmp = pool.dag.getEpochRef(target.blck, target.slot.epoch, false)
@@ -641,6 +656,9 @@ proc validateAggregate*(
       return checkedReject("Attestation: committee index not within expected range")
     idx.get()
 
+  # [REJECT] aggregate_and_proof.selection_proof selects the validator as an
+  # aggregator for the slot -- i.e. is_aggregator(state, aggregate.data.slot,
+  # aggregate.data.index, aggregate_and_proof.selection_proof) returns True.
   if not is_aggregator(
       epochRef, slot, committee_index, aggregate_and_proof.selection_proof):
     return checkedReject("Aggregate: incorrect aggregator")
diff --git a/tests/test_attestation_pool.nim b/tests/test_attestation_pool.nim
index e77890581..ce41aa2a8 100644
--- a/tests/test_attestation_pool.nim
+++ b/tests/test_attestation_pool.nim
@@ -202,6 +202,9 @@ suite "Attestation pool processing" & preset():
     att0.combine(att2)
     att1.combine(att2)
 
+    check:
+      not pool[].covers(att0.data, att0.aggregation_bits)
+
     pool[].addAttestation(
       att0, @[bc0[0], bc0[2]], att0.loadSig, att0.data.slot.start_beacon_time)
     pool[].addAttestation(
@@ -214,6 +217,7 @@ suite "Attestation pool processing" & preset():
         info, {}).isOk()
 
     check:
+      pool[].covers(att0.data, att0.aggregation_bits)
       pool[].getAttestationsForBlock(state.data, cache).len() == 2
       # Can get either aggregate here, random!
       pool[].getAggregatedAttestation(1.Slot, 0.CommitteeIndex).isSome()