Reset A A Font size: Print

United States Court of Appeals,Second Circuit.

UNITED STATES of America, Appellee, v. Curtis TAYLOR, Antonio Rosario, aka Chickee, Samuel Vasquez, aka Rock, Defendants–Appellants.

Nos. 11–2201, 11–2426, 11–2639.

Decided: May 23, 2014

ROBERT A. KATZMANN, Chief Judge, DENNIS JACOBS, JOSÉ A. CABRANES, ROSEMARY S. POOLER, REENA RAGGI, RICHARD C. WESLEY, PETER W. HALL, DEBRA ANN LIVINGSTON, GERARD E. LYNCH, DENNY CHIN, RAYMOND J. LOHIER, JR., SUSAN L. CARNEY, and CHRISTOPHER F. DRONEY, Circuit Judges. Christopher D. Frey, Brent S. Wible, and Michael A. Levy, Assistant United States Attorneys, for Preet Bharara, United States Attorney for the Southern District of New York, New York, NY. Kelley J. Sharkey, Brooklyn, NY, for Defendant–Appellant Taylor. Jillian S. Harrington, Monroe, Township, NJ, for Defendant–Appellant Rosario. Colleen P. Cassidy, Assistant Federal Public Defender, Federal Defenders of New York, Inc., New York, NY, for Defendant–Appellant Vasquez.


Following disposition of this appeal on March 4, 2014, an active judge of the Court requested a poll on whether to rehear the case en banc. A poll having been conducted and there being no majority favoring en banc review, rehearing en banc is hereby DENIED.

I write as one of the majority of active judges who voted to deny rehearing en banc in this case. I write only for myself. I carefully studied the memoranda circulated by my colleagues who voted in favor of rehearing en banc, which presented arguments similar to those set forth in Judge Raggi's dissent from the denial of en banc review. I did not vote against rehearing en banc because of some imagined distaste for en banc proceedings or because I thought that the issues presented here are so important we should expedite the Supreme Court's consideration of this case. Instead, I voted to deny rehearing en banc because I believe that the panel's decision in this case is substantively correct.

I respectfully join, without qualification, in Judge Raggi's forceful opinion. The dissenters having failed to persuade a majority of the active judges to rehear this appeal, our concerns necessarily now rest in the hands of our highest court. I write separately, and in my name alone, for the sole purpose of recalling some salient aspects of en banc practice in the Second Circuit.1

As a result of our decision not to rehear this case before the full court of active judges, by a vote of seven to six, one can know for certain only one thing: Judge Raggi's opinion dissenting from the denial of en banc review is, by definition, an expression of the views of the six subscribing judges that the panel's resolution of this case presents legal issues of exceptional importance, and defies not only our own case law, but controlling Supreme Court precedent. In contrast, the order itself denying rehearing without elaboration may, or may not, reflect the substantive views of particular judges in the seven-judge majority voting against rehearing.2 This is so because, even when the criteria enumerated in Rule 35 are satisfied—when “(1) en banc consideration is necessary to secure or maintain uniformity of the court's decisions; or (2) the proceeding involves a question of exceptional importance”3 —the active circuit judges nonetheless exercise considerable discretion to vote against en banc rehearing.4

In light of how judges of the Second Circuit have historically exercised such discretion, the decision not to convene the en banc court does not necessarily mean that a case either lacks significance or was correctly decided. Indeed, the contrary may be true. An oft-cited justification for voting against rehearing, perhaps counterintuitively, is that the case is “too important to en banc.”5 This view was apparently first stated in published case law in 1973 by Judge Irving R. Kaufman, who voted against en banc consideration of a panel decision in order to “wisely speed this case on its way to the Supreme Court as an exercise of sound, prudent and resourceful judicial administration.”6 This view has perdured,7 leaving open the possibility that some judges in the majority in any particular case may have voted against en banc rehearing precisely because of the importance of the legal questions at issue.

Other reasons for voting against rehearing that may not be related to the case's merits can be grouped under what Judge Jon O. Newman called the “virtues of restraint.”8 Such “virtues” include judicial economy,9 collegiality,10 and what Chief Judge Robert A. Katzmann more recently described as “our Circuit's longstanding tradition of general deference to panel adjudication-a tradition which holds whether or not the judges of the Court agree with the panel's disposition of the matter before it.”11

Accordingly, a reader should not attempt to ascertain the substantive views of particular judges in the majority (or even, in some circumstances, the unsuccessful minority) from a decision not to rehear a case en banc. Nor should a reader accord any extra weight to a panel opinion in light of such a decision, inasmuch as the order denying rehearing may only reflect, for some judges, a general aversion to en banc rehearings or faith in the Supreme Court to remedy any major legal errors.

In sum, all one can know for certain about a vote like this one is that seven active circuit judges did not wish to rehear this case, while the six other active circuit judges strongly believed that the panel opinion presented multiple legal errors of exceptional importance warranting correction.

In vacating convictions in this Hobbs Act robbery case, a panel of the court reaches the paradoxical conclusion that a defendant who acted knowingly and voluntarily in waiving his Miranda rights could not have acted knowingly and voluntarily in responding to ensuing police questions. Why? Not because of any abusive police interrogation tactics—the panel concedes there were none—but, rather, because defendant was so sleepy that he occasionally dozed off during the interview.1 The panel then concludes that the “coercive” effect of the first interrogation carried over to a second the following day, even though defendant himself solicited the second interview, re-waived his Miranda rights, and was awake throughout.

These conclusions defy not only common sense but also controlling precedent, notably, Dickerson v. United States, 530 U.S. 428 (2000), Colorado v. Connelly, 479 U.S. 157 (1986), and Oregon v. Elstad, 470 U.S. 298 (1985). This prompts six of the thirteen active judges in this circuit to seek review of this case en banc. That review should also extend to the panel's identification of error under Bruton v. United States, 391 U.S. 123 (1968), in the admission of purportedly inadequately redacted versions of the confessions at a joint trial. In fact, the redactions replace co-defendants' names with neutral substitutes solicited by defendants and previously approved by this and other federal courts. Further, co-defendants can be linked to the redacted statements only by looking to other evidence, circumstances that cannot demonstrate Bruton error. See Richardson v. Marsh, 481 U.S. 200,206–09 (1987).

Thus, to the extent the court today declines to grant en banc review, I respectfully dissent from that decision.

A. A Preliminary Observation About “Facts”

Throughout this opinion, I assume readers' familiarity with the panel opinions, see United States v. Taylor (“Taylor I ”), 736 F.3d 661 (2d Cir.2013), superseded by United States v. Taylor (“Taylor II ”), 745 F.3d 15 (2d Cir.2014), and I generally detail relevant facts only as necessary to explain why en banc review is warranted. At the outset, however, it is important to note that the panel's troubling legal analysis rests on a suspect factual characterization of defendant Curtis Taylor's condition at the time of the challenged confessions.

Notably, the panel asserts, based on its own reading of the record, that Taylor was “in and out of consciousness,” “in a trance or a stupor,” “largely stupefied,” and “unable to focus” even when awake during his first interrogation. Taylor II, 745 F .3d at 20, 25. Such characterizations are findings of fact and, as such, outside the purview of an appellate court. They are, moreover, at odds with factual findings of the district court that were not clearly erroneous. See Ornelas v. United States, 517 U .S. 690, 699 (1996).

After an evidentiary hearing, the district court expressly found that, “during the questioning,” Taylor was “sufficiently lucid,” “awake,” and “competent” to exercise his constitutional rights. Suppression Hr'g Tr. (“Tr.”) 387:23–388:14, Supplemental App'x (“S.A.”) 387–88. In making these findings, the district court credited interviewing officers who testified that, during the first interrogation, Taylor was “coherent,” “fluid,” “knew what was going on,” and never asked for questioning to cease. Id. 18:25–20:3, S.A. 18–20. While acknowledging that Taylor may have “nodd[ed] off” two or three times during the initial two-to-three hour interview, an FBI agent explained that, upon verbal prompting, Taylor was able to focus on the questions posed: he “would respond that he knew what he was being asked and he would repeat the questions back to us to show that he was understanding what was being asked of him and knew what was going on.” Id. 45:7–21, S .A. 45. As to the next day's interview—which Taylor himself requested—the agent testified that Taylor remained awake throughout, “appeared fine,” and participated in a “lucid give and take”; indeed, “[h]e was probing with information that he wanted to clarify, and that led to my [i.e., the agent's] questions to him. I didn't note any confusion ․ aside from what he wanted to clarify.” Id. 216:17–21, S.A. 216.

To support its own contrary factual assessment, the panel highlights other hearing evidence suggesting that at different times on the dates in question, Taylor fell asleep easily or experienced difficulties with mental focus and verbal expression.2 But the task of weighing any competing evidence was committed to the district court, not the panel. Cf. United States v. Mendenhall, 446 U.S. 544, 557 (1980) (“[B]ecause the trial court's finding [that defendant had acted voluntarily in accompanying police to station] was sustained by the record, the Court of Appeals was mistaken in substituting for that finding its view of the evidence.”); United States v. Isiofia, 370 F.3d 226, 232 (2d Cir.2004) (upholding district court's finding that consent to search was not voluntary “even though ․ had we been sitting as the trier of fact, we would have weighed the evidence differently” (internal quotation marks and alterations omitted)). Indeed, absent identification of clear factual error—which the panel does not make here—it cannot substitute its own factual assessment of Taylor's condition at the time of his confessions for that of the district court. See United States v. Khalil, 214 F.3d 111, 121–22 (2d Cir.2000) (deferring to district court's voluntariness finding even though circumstances surrounding confession might have supported contrary inference); see generally Anderson v. Bessemer City, 470 U.S. 564, 574 (1985) (“Where there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous.”).

To the extent the panel does so, such a departure from long-standing precedent might itself warrant correction en banc. Here, however, it is sufficient to note this factfinding concern before discussing the relevant legal issues, which warrant en banc review even under the panel's own assessment of the facts.

B. Legal Errors Warranting En Banc Review

1. The Purported Involuntariness of Taylor's First Confession

The panel holds that for law enforcement officers to have questioned Taylor when he was intermittently falling asleep so overbore his will as to render any admissions constitutionally involuntary. See Taylor II, 745 F.3d at 25. But as the panel itself acknowledges, Taylor knowingly and voluntarily waived his Miranda rights at the start of the interview and was not subjected to any abusive interrogation tactics. Id. at 23, 25 & n.1. In these circumstances, the panel's identification of constitutional error cannot be reconciled with Dickerson v. United States, 530 U.S. 428, and Colorado v. Connelly, 479 U.S. 157.

a. Dickerson's Application to this Case

Whether police coerced a confession by improperly taking advantage of a defendant's impaired condition is a fact-intensive inquiry on which we usually defer to the district court.3 Here, the panel defers to the district court's finding that Taylor knowingly and voluntarily waived his Miranda rights before his first confession, see Taylor II, 745 F.3d at 23, but then departs from the district court in holding Taylor's ensuing confession involuntary as a matter of law. In support, the panel cites Dickerson v. United States, 530 U.S. 428, for the conclusion that a Miranda waiver does not guarantee that subsequent statements were constitutionally voluntary. See Taylor II, 745 F.3d at 23.

To be sure, Dickerson instructs that “[t]he requirement that Miranda warnings be given does not, of course, dispense with the voluntariness inquiry.” 530 U.S. at 444. But this text is immediately followed by a caution that the panel fails to acknowledge and does not heed: that “cases in which a defendant can make a colorable argument that a self-incriminating statement was ‘compelled’ despite the fact that the law enforcement authorities adhered to the dictates of Miranda are rare.” Id. (internal quotation marks and alterations omitted); accord In re Terrorist Bombings of U.S. Embassies in E. Afr., 552 F.3d 177, 212 (2d Cir.2008). Implicit in Dickerson's rarity admonition is the recognition that, among the totality of circumstances that determine voluntariness, Miranda waivers bear considerable weight. See United States v. Williams, 681 F.3d 35, 45 (2d Cir.2012) (stating that suspect's knowing and voluntary waiver of rights is “ ‘highly probative’ of voluntariness” (quoting Oregon v. Elstad, 470 U.S. at 318)); see also McNeil v. Wisconsin, 501 U.S. 171, 181 (1991) (observing that “[a]dmissions of guilt resulting from valid Miranda waivers are more than merely desirable; they are essential to society's compelling interest in finding, convicting, and punishing those who violate the law” (internal quotation marks omitted)).

While the panel accepts Taylor's valid Miranda waiver, it accords little, if any, weight to the waiver in assessing the voluntariness of his ensuing confession. Rather, the panel focuses almost exclusively on Taylor's sleepiness during his initial police interview and concludes therefrom that he “was unable to summon the will to make a knowing and voluntary decision” about speaking to the police. Taylor II, 745 F.3d at 24. But this effectively misses Dickerson 's point. The purportedly sleepy Taylor had demonstrated himself able to make just such a “knowing and voluntary decision” moments earlier when, at the start of the interview, he validly waived his Miranda rights.

Indeed, this record provides no basis for the panel's decision to accept the district court's finding that Taylor was sufficiently competent to waive his Miranda rights but to reject the district court's same finding of competency with respect to his ensuing confession. Certainly, the district court's competency finding cannot fairly be construed to apply only to the moment that Taylor executed his written Miranda waiver. The record indicates that the district court viewed the question in dispute to be whether Taylor's condition throughout the police interrogation cast doubt on the continued validity of his waiver of rights. Thus, it found that Taylor was “sufficiently lucid during the questioning that his waiver of Miranda rights was knowing and voluntary.” Tr. 387:23–25, S.A. 387 (emphasis added). Moreover, unlike the defendant in Mincey v. Arizona, 437 U.S. 385, 399–400 (1978), Taylor never indicated that he wished questioning to stop. To the contrary, on the few occasions when he was prompted to focus on a question, Taylor stated that he understood what was being asked, repeated the question to demonstrate his comprehension, and then provided a response. Further, the detailed and cogent nature of Taylor's confessions not only supports the district court's finding that he was sufficiently awake and lucid to participate voluntarily in the post-arrest interview but also precludes the panel's contrary assessment of “stupor” or “trance.”

In the absence of any finding of clear error in the district court's factual determination (which, as noted, the panel does not make here), I respectfully submit it is not possible, consistent with Dickerson, for a reviewing court to conclude as a matter of law that this is one of the “rare” cases in which admissions made after a valid Miranda waiver are, nevertheless, constitutionally involuntary. Thus, en banc review is warranted to ensure our court's adherence to Dickerson.

b. Connelly's Application to this Case

Insofar as Taylor's sleepiness is the singular reason for the panel denominating his initial interrogation as coercive, a further concern arises with respect to this court's faithful adherence to Colorado v. Connelly, 479 U.S. 157. While a defendant's mental state may be relevant to assessing the voluntariness of a challenged confession, see Withrow v. Williams, 507 U.S. 680, 693 (1993), Connelly instructs that it cannot, “by itself and apart from its relation to official coercion ․ ever dispose of the inquiry into constitutional ‘voluntariness,’ “ 479 U.S. at 164; see id. at 166 (stating that defendant has no constitutional right “to confess to his crime only when totally rational and properly motivated”). Rather, Connelly holds that “coercive police activity is a necessary predicate ” to any finding of constitutional involuntariness. 479 U.S. at 167 (emphasis added); see id. at 163–64 (emphasizing “crucial element of police overreaching” that characterizes coercive action). Following Connelly, this court has recognized that a defendant's “mental state does not become part of the calculus for the suppression of evidence unless there is an allegation that agents ․ engaged in some type of coercion.” United States v. Salameh, 152 F.3d 88, 117 (2d Cir.1998) (emphasis added).

Here, the panel itself acknowledges that “[t]he conditions in which Taylor was questioned do not appear to have been abusive.” Taylor II, 745 F.3d at 25; see also id. at 25 n.1. Moreover, by accepting the district court's finding of a valid Miranda waiver, the panel effectively recognizes the removal of any coercion inherent in the fact that Taylor's questioning was custodial. See Miranda v. Arizona, 384 U.S. 436, 467–68 (1966); see generally Oregon v. Elstad, 470 U.S. at 310–11 (holding that “careful and thorough administration of Miranda warnings serves to cure the condition” that renders inadmissible unwarned confession during custodial interrogation). In these circumstances, there is no record basis consistent with Connelly for holding Taylor's first confession constitutionally involuntary based on the fact that he was questioned while sleepy.

To be sure, if a defendant's sleepiness were itself the product of deliberate police action, that action would satisfy the overreaching prerequisite. See Ashcraft v. Tennessee, 322 U.S. 143, 153–54 (1944) (holding 36–hour questioning coercive); accord United States ex rel. Burns v. LaVallee, 436 F.2d 1352, 1355–56 (2d Cir.1970) (holding coercive 18–hour questioning where defendant had been without sleep for 30 hours). The same conclusion would obtain if police persisted in questioning a sleep-deprived person who repeatedly asked that questioning cease. See Mincey v. Arizona, 437 U.S. at 399–402. But to the extent the panel cites Mincey and LaVallee to support its conclusion that “[c]ontinued questioning of a sleep-deprived suspect can be coercive, depending on the circumstances,” Taylor II, 745 F.3d at 25, the critical “circumstance,” present in those cases and missing here, is official overreaching reflected in some action apart from the mere questioning of a sleepy person.

Indeed, Connelly observed that “all” prior Supreme Court decisions holding confessions involuntary “contained a substantial element of coercive police conduct.” 479 U.S. at 163–64 (emphasis added). In discussing the type of police overreaching with respect to impaired persons that rises to this level, Connelly makes plain that it contemplates something more than questioning. Referencing Blackburn v. Alabama, 361 U.S. 199 (1960), Connelly explains that police there exploited a possibly insane defendant's mental weakness with “coercive tactics” that included “eight- to nine-hour sustained interrogation in a tiny room.” 479 U.S. at 164–65 (internal quotation marks omitted).4 Connelly cites Townsend v. Sain, 372 U.S. 293 (1963), as “present[ing] a similar instance of police wrongdoing,” in that a police physician had there given the defendant a truth-serum drug, a fact known to police interrogators when they elicited defendant's confession. 479 U.S. at 165. No analogous abuse having attended officers' questioning of Taylor, the panel's identification of constitutionally impermissible “coercion” cannot be reconciled with Connelly.

Compounding the panel's Connelly error is its pronouncement that “there is little difference in effect between sleep deprivation as a technique and the relentless questioning of a person who is obviously unable to focus or stay awake for some other reason.” Taylor II, 745 F.3d at 25.5 By suggesting that the relevant constitutional focus is on the effect rather than the means of interrogation, the panel elides the “necessary predicate” demanded by Connelly: “coercive police activity.” 479 U.S. at 167. Indeed, Connelly appears to have rejected an effects-focused test for constitutional involuntariness by distinguishing between circumstances that might render an incapacitated defendant's confession “unreliable”—a matter that Connelly concludes should generally be governed by a forum's evidentiary rules—and circumstances demonstrating “coercive” police activity so “fundamental[ly] unfair[ ]” as to violate due process. Id. (internal quotation marks omitted).

In sum, where, as here, a defendant knowingly and voluntarily waived his Miranda rights, and where police thereafter questioned him without employing any abusive tactics, a reviewing court cannot conclude, consistent with Connelly and contrary to the factual findings of the district court, that simply questioning Taylor while he was sleepy was fundamentally unfair, much less that it overbore his will, so as to render responses constitutionally involuntary. To the extent the panel so held, see Taylor II, 745 F.3d at 25, we should correct its Connelly error.

2. The Purported Continuing Involuntariness of Taylor's Second Confession

Even if one were to assume arguendo that Taylor's first confession was coerced, the panel's conclusion that his second confession was thus also involuntary warrants en banc review because it (a) fails to apply—and improperly narrows—the totality-of-the-circumstances review that determines the ultimate question of continuing compulsion; and (b) fails to accord any weight to other circumstances more relevant to the issue of continued coercion, while mistakenly grounding a presumption of continued coercion in Taylor's first confession having “ ‘let the cat out of the bag.’ “ Taylor II, 745 F.3d at 25 (quoting United States v. Bayer, 331 U.S. 532, 540 (1947)).

a. Continuing Coercion Must Be Assessed by Reference to the Totality of the Circumstances

The panel limits its continuing coercion inquiry to three factors identified in Oregon v. Elstad: the place of interrogation, the time between confessions, and the identity of the interrogators. See 470 U.S. at 310. Nevertheless, it concludes that because less than 24 hours had passed between Taylor's confessions and because at least one common agent participated in both interrogations, the second confession was burdened “with a ‘presumption of compulsion’ “ requiring suppression. Taylor II, 745 F.3d at 26 (citation omitted).6

The panel errs in so limiting its continuing coercion inquiry. As Elstad states, its identified factors “bear on” the constitutional question of whether compulsion prompting a first confession “has carried over into the second”; the factors do not cabin the taint inquiry or necessarily determine it. 470 U.S. at 310. The question of continuing coercion, like the question of voluntariness generally, depends on the totality of the circumstances. See Lyons v. Oklahoma, 322 U.S. 596, 602 (1944) (stating that whether “confessions subsequently given are themselves voluntary depends on the inferences as to the continuing effect of the coercive practices which may fairly be drawn from the surrounding circumstances”); see also id. at 603 (“The admissibility of the later confession depends upon the same test—is it voluntary.”); Tankleff v. Senkowski, 135 F.3d 235, 244–45 (2d Cir.1998) (holding that courts must look to “totality of the circumstances” in deciding whether second confession following coerced statement must also be suppressed).

b. Relevant, Yet Disregarded, Considerations

By focusing exclusively on the Elstad factors, the panel in fact overlooks the two circumstances bearing most directly on, and ultimately belying, continued coercion in this case: (1) the limited continuing effect of sleepiness as a “coercive” factor, and (2) Taylor's own request for a second police interview.

(1) The Limited Continuing Effect of Sleepiness

In Tankleff v. Senkowski, this court recognized that the particular coercion informing a first confession is properly considered to determine whether it irredeemably taints a second confession. See 135 F.3d at 244–45. This is because coercion is not all of a kind. Where it is achieved through physical force, the presumption of continuing coercion may be strongest, not only because of lingering pain but also because of feared repetition. Cf. Lyons v. Oklahoma, 322 U.S. at 604 (concluding that taint from use of force had dissipated because defendant had no reason to fear mistreatment in second interrogation). Similarly, coercive deceit that is not corrected can have a continuing effect because a defendant may be relying on the same misrepresentations in making his second admission as he did in making his first. See United States v. Anderson, 929 F.2d 96, 102 (2d Cir.1991) (relying on fact that agent “made no effort to dispel the original threat” but, rather, “reaffirmed [other agent's] earlier coercive statements” in holding second statement tainted).7

The “coercion” at issue here is of a very different sort. Given the panel's acknowledgment that no abusive questioning tactics were employed, its only reason for viewing the initial interview informing Taylor's first confession as “coercive” is that the questioning was pursued while he was intermittently dozing off. Even assuming that such a determination of coercion could be squared with Connelly, the coercive reach of such conduct—by contrast to physical abuse or deceit—is not long. Indeed, I do not see how it can be presumed to continue beyond the sleepiness that supports it.

The panel acknowledges agent testimony that Taylor never fell asleep during his second interrogation. See Taylor II, 745 F.3d at 26. But rather than recognize this as a strong factor militating against continuing coercion from the prior day's sleepiness, the panel appears to question the testimony's reliability by pointing to “uncontradicted testimony” of Taylor's continued mental impairment on the day of the second interrogation. Id.

As noted supra at [4 & n.2], that testimony pertains to observations of Taylor at times other than during the interview when he made his second confession. As to that 20–minute period, the “uncontradicted testimony” of the credited interviewing agent was that Taylor “appeared fine” throughout and participated in a “lucid give and take” respecting information that he wanted to clarify. Tr. 216:17–21, S.A. 216. Thus, whatever inferences a factfinder might have drawn from Taylor's behavior at times distinct from the second interview, one thing is clear: a reviewing court cannot itself weigh that evidence and conclude therefrom that Taylor was not awake and lucid when he made his second confession—at least not without rejecting the district court's express finding that the agents who so testified to his condition were credible. See, e.g., United States v. Iodice, 525 F.3d 179, 185 (2d Cir.2008) (emphasizing “particularly strong deference” owed district court's credibility determinations).

Accordingly, even if—despite Dickerson and Connelly—Taylor's sleepiness during the first confession had rendered it coercive to question him in that condition, his ability to stay awake throughout the second interview is a factor that strongly weighs against a conclusion of continuing coercion.

(2) Taylor's Initiation of the Second Police Interview

The panel also accords little if any weight to the fact that Taylor himself sought the second interview with law enforcement officials. Where a defendant thus seeks out the authorities to initiate a second interview, and where a prior incriminating statement is deemed coercive only because the defendant was sleepy during that questioning—not because abusive interrogation tactics were employed—I submit that a presumption of continuing coercion cannot attach to that second interview solicited by the defendant, for which he again waived Miranda rights, and throughout which he was awake.

No different conclusion is warranted by United States v. Bayer, which the panel cites to explain that the reason a presumption of continuing compulsion arises from a coerced confession is “because, ‘after an accused has once let the cat out of the bag by confessing, no matter what the inducement, he is never thereafter free of the psychological and practical disadvantages of having confessed.’ “ Taylor II, 745 F.3d at 25 (quoting United States v. Bayer, 331 U.S. at 540). When this language of Justice Jackson's in Bayer is read in context, however, it makes a quite different point, namely, that although a first confession lets the “cat out of the bag,” that does not preclude admission of a second confession at trial, particularly where the second confession is made after the original conditions of coercion have been removed:

Of course, after an accused has once let the cat out of the bag by confessing, no matter what the inducement, he is never thereafter free of the psychological and practical disadvantages of having confessed. He can never get the cat back in the bag. The secret is out for good. In such a sense, a later confession always may be looked upon as fruit of the first. But this Court has never gone so far as to hold that making a confession under circumstances which preclude its use, perpetually disables the confessor from making a usable one after those conditions have been removed.

United States v. Bayer, 331 U.S. at 540–41 (emphasis added); accord Tanner v. Vincent, 541 F.2d 932, 937 (2d Cir.1976) (instructing that although first confession “let the cat out of the bag,” that is “only one factor” bearing on voluntariness of later statement; existence of first confession “should not, in itself, always be fatal” to use of second admission at trial (internal quotation marks omitted)).

In sum, Taylor's initiation of the second interview, together with the agents' re-administration of Miranda rights, Taylor's waiver of those rights, and his ability to remain awake and lucid throughout the 20–minute interview, compel a conclusion that, whatever, if any, coercion attached to Taylor's first confession because of his sleepiness, it did not “carr[y] over” into the second confession. Oregon v. Elstad, 470 U.S. at 310. To the extent the panel concluded otherwise by looking only to intervening time and common participants, factors less relevant to the continuing coercion inquiry in this case, en banc review is necessary to reaffirm that totality-of-the-circumstances is the appropriate standard of review.

3. Vacatur of Co–Defendants' Convictions for Purportedly Inadequate Redaction of Taylor's Confessions

In its first opinion, the panel identified no concern with how Taylor's confessions had been redacted for use at a joint trial with co-defendants Antonio Rosario and Samuel Vasquez. Rather, it vacated these co-defendants' convictions based on a risk that the jury had ignored instructions to consider the confessions only as to Taylor. See Taylor I, 736 F.3d at 673–74. Upon reconsideration, the panel abandons this position. Nevertheless, it persists in its vacatur decision, concluding that the redacted statements contained “obvious” indications of alteration from which the jury would “immediately infer” that Taylor had inculpated Rosario and Vasquez. Taylor II, 745 F.3d at 27–30. This new holding warrants en banc review because it marks an unprecedented extension of Bruton v. United States, 391 U.S. 123, to neutral substitution language previously approved by this court and solicited by co-defendants, and depends on viewing the redacted statements together with other trial evidence contrary to Richardson v. Marsh, 481 U.S. 200.

In Bruton, the Supreme Court held that admission at a joint trial of a nontestifying defendant's unredacted confession inculpating a co-defendant violates the latter's Confrontation Clause rights because of the great risk that a jury cannot follow instructions to consider such evidence against only the confessing defendant. See 391 U.S. at 135–36. This rule, however, is a “narrow exception” to the “almost invariable” assumption that juries follow instructions and one that the Supreme Court has carefully limited. Richardson v. Marsh, 481 U.S. at 206–09 (identifying no Bruton error where redactions eliminated co-defendant's existence because confession was not “incriminating on its face, [but] became so only when linked with evidence introduced later at trial”). Indeed, the Supreme Court has extended Bruton to preclude admission of a redacted confession only where a co-defendant's name was replaced with “a blank space or the word ‘deleted,’ “ i.e., “obvious indications of alteration” that “so closely resemble Bruton 's unredacted statements” as to “refer[ ] directly to the ‘existence’ of the nonconfessing defendant.” Gray v. Maryland, 523 U.S. 185,188, 192 (1998).

From these precedents, our court has determined that admission of a redacted confession violates Bruton if the statement (a) contains “obvious indications of alteration” that signal to the jury that the original statement “contained actual names”; or, (b) even if viewed in isolation as the “very first item introduced at trial,” “immediately” inculpates the co-defendant in the charged crime. United States v. Jass, 569 F.3d 47, 61–62 (2d Cir.2009) (internal quotation marks omitted). Here, the panel's analysis of both factors raises concerns.

a. Obvious Indications of Alteration

In concluding that Taylor's redacted statements displayed “obvious indications” of alteration, the panel holds—for the first time by our court—that neutral redactions were insufficient to avoid Bruton error.8 In support, it points to dictum in United States v. Jass, which in rejecting a Bruton challenge to a redacted confession in that case, left open the possibility that neutral redactions might be so ham-handed as to result in constitutional injury, see 569 F.3d at 61–62 (using example of direct quote redacted to read, “Look, other person, we have to get out of here' “ (emphasis added)). See Taylor II, 745 F.3d at 29.

Here, the panel does not—and cannot—identify any substitutions in Taylor's confessions as awkward as that hypothesized in Jass.9 Nevertheless, the panel concludes that the employed neutral substitutions—including the words “persons” and “individuals”—produced “stilted circumlocutions” that, with the identification of confederate Luana Miller by name, demonstrated such obvious redaction as to violate Bruton. Taylor II, 745 F.3d at 29. The conclusion is flawed in several respects.

First, the panel's reliance on the inclusion of Miller's name together with neutral substitutions for co-defendants' own as a basis for identifying Bruton error conflicts with United States v. Tutino, 883 F.2d 1125 (2d Cir.1989). There, we upheld the admission of a redacted confession that replaced two co-conspirators' names with “others,” “other people,” and “another person,” but still identified confederate Vince Cafaro, observing that “the jury never knew that Tutino's original statement named names, and could easily have concluded that Tutino did not want to reveal the identity of his coconspirators.” Id. at 1135.10

To be sure, in Tutino, the confessing defendant knew that Cafaro was cooperating with law enforcement authorities and, thus, already known to them. Id. at 1137. But there are any number of reasons why a defendant might name one confederate (a grudge against the named party, blaming the named party for the declarant's own predicament, an expectation that the named party will avoid apprehension or be treated leniently, etc.) and not another (loyalty, family relations, fear, etc.). Here, Taylor might well have named Miller because he knew she was unmasked while inside the robbed pharmacy and, thus, most likely to be identified by the police, and yet tried to shield the identities of masked gunman Rosario and getaway driver Vasquez.11 Indeed, Rosario's counsel himself argued to the district court that the jury could readily conclude that Taylor had chosen to shield his co-defendants while, nevertheless, naming Miller. See S.A. 409:2–12. Moreover, neither Rosario's nor Vasquez's counsel ever objected to Miller being named in the redacted confessions or sought redaction. In such circumstances, there is no basis for the panel to depart from Tutino.12

Second, as to the panel's concern with “stilted circumlocutions,” the purported awkwardness of the employed redactions is not only not akin to that hypothesized in Jass. It also does not “so closely resemble Bruton 's unredacted statements” as did the use of a blank space or the word “deleted” in Gray, 523 U.S. at 192. Thus, the panel provides no good reason to depart from courts' frequent approval of neutral substitutes such as “person” and “individual” for co-defendants' names. See, e.g., United States v. Jass, 569 F.3d at 61 (approving use of “another person”); United States v. Tutino, 883 F.2d at 1135 (approving “others,” “other people,” and “another person”); see also United States v. Green, 648 F.3d 569, 575–76 (7th Cir.2011) (upholding use of “straw buyer” because it is “similar to an anonymous reference such as ‘another person’ or ‘an individual’ ”). Much less does the panel provide any guidance for district courts as to when such neutral substitutions will be held so “stilted” as to manifest constitutional error.

Third, co-defendants' own actions hardly support the panel's departure from precedent to extend Bruton to the neutral redactions in this case. Although co-defendants objected generally to the admission of Taylor's confessions, they did not object to the particular language ultimately employed by the district court. To the contrary, they solicited the use of such gender-neutral substitutes, complaining that the government's suggested substitution of the more colloquial “guys” implied male confederates. See S.A. 404:23–410:2. Indeed, when the district court itself expressed concern that co-defendants' requests would lead to an unnatural syntax, Rosario's counsel insisted that what he was proposing would “seem more realistic to the jury” and not “awkward.” S.A. 409:10–25; see also id. 406:20–24 (maintaining that proposed gender-neutral substitutions would make statements “less awkward,” “more readable,” and “less obvious[ly] redacted”). Further, Rosario's counsel argued that, if there were any awkwardness, the jury might well attribute it to Taylor's deliberate efforts not to name his confederates, see S.A. 409:2–12–the very conclusion that supported our rejection of a Bruton challenge in Tutino, see 883 F.2d at 1135. Such defense conduct might well qualify as true waiver of any complaint about stilted circumlocutions. See United States v. Quinones, 511 F.3d 289, 321 (2d Cir.2007) (stating that “true waiver applies with even more force when ․ defendants not only failed to object to what they now describe as error, but they actively solicited it”).

On this record, there is no sound basis in law to conclude that redacted confessions employing neutral substitutes, urged by co-defendants and previously approved by this court, caused constitutional injury requiring vacatur of co-defendants' convictions.

b. Immediate Reference to Co–Defendants

Nor does established precedent support the panel's conclusion that, because the “unnamed persons [in the redacted statements] correspond by number (two) and by role to the pair of co-defendants,” the redacted confessions necessarily referred to Rosario and Vasquez. Taylor II, 745 F.3d at 29.

In Jass, we squarely rejected such a numerosity argument, upholding the replacement of the single female co-defendant's name with the singular “another person”—which, given the particulars of the confession, likely referred to a female—because “a jury would have had to refer to other trial evidence to link [that co-defendant] to the redacted statement.” 569 F.3d at 62–63. In evaluating Bruton challenges to redacted confessions, a reviewing court properly considers the confessions in isolation, without regard to other evidence. See Richardson v. Marsh, 481 U.S. at 208 (holding that Bruton does not categorically exclude confession that is incriminating “only when linked with evidence introduced later at trial”); cf. Gray v. Maryland, 523 U.S. at 196 (holding redaction inadequate because jury could “immediately” infer that declarant inculpated codefendant, “even were the confession the very first item introduced at trial”). This reasoning applies equally to role as to numerosity.

Considering the redacted confessions here in isolation, the jury would not immediately know that Taylor had assigned the two confederate roles described therein to co-defendants Rosario and Vasquez. Such an inference depended on other evidence, notably, Miller's testimony. Following Richardson, this court has consistently declined to identify Bruton error where statements thus inculpate co-defendants only “when placed in context with other testimony.” United States v. Lung Fong Chen, 393 F.3d 139, 150 (2d. Cir.2004); accord In re Terrorist Bombings of U.S. Embassies in E. Afr., 552 F.3d 93, 134 (2d Cir.2008). Insofar as the panel decision now departs from this precedent, it warrants correction en banc.

* * *

In sum, en banc review is required in this case,

1. to ensure our court's adherence to Dickerson v. United States, 530 U.S. 428, and Colorado v. Connelly, 479 U.S. 157, in evaluating the voluntariness of a confession made by an impaired defendant after a valid waiver of Miranda rights and in response to non-abusive questioning;

2. to clarify that, consistent with Oregon v. Elstad, 470 U.S. 298, Lyons v. Oklahoma, 322 U.S. 596, and Tankleff v. Senkowski, 135 F.3d 235, the assessment of a claim of continuing coercion is not cabined by the trio of illustrative factors identified in Elstad, but properly extends to the totality of circumstances—which can include the type of coercion initially employed and defendant's own initiation of the second police interview; and

3. to maintain our Bruton jurisprudence consistent with the neutral redaction principles articulated in Richardson v. Marsh, 481 U.S. 200, Gray v. Maryland, 523 U.S. 185, United States v. Jass, 569 F.3d 47, and United States v. Tutino, 883 F.2d 1125.

Because of the significance of these legal concerns to our jurisprudence generally and because proper application of these precedents to this case would result in affirmance of all three defendants' challenged convictions rather than the vacatur ordered by the panel, I respectfully dissent from the denial of rehearing en banc.


1.  It seems worthwhile to explain again the variation in the number of en banc dispositions between the Second Circuit and our sister courts of appeal, perhaps tracing back to Judge Learned Hand's promise that he would never vote to convene an en banc court. Gerald Gunther, Learned Hand: The Man and the Judge 515–16 (1994); see also Wilfred Feinberg, The Office of Chief Judge of a Federal Court of Appeals, 53 Fordham L.Rev. 369, 376 (1984) (“The tradition in the Second Circuit, a tradition that goes back to Learned Hand, is that in bancs are not encouraged.”). Our Court hears the fewest cases en banc of any circuit by a substantial margin, both in absolute terms and when considering the relative size of our docket. See Federal Bar Council, Second Circuit Courts Committee, En banc Practices in the Second Circuit: Time for a Change? 6 (July 2011). As the Federal Bar Council has observed, “[t]he vast difference ․ indicates that something different is happening when the judges of the Second Circuit consider whether to grant en banc review.” Id. at 22.

2.  Indeed, nothing about the merits of a case is revealed in the standard order denying rehearing, which, as here, states: “Following disposition of this appeal on [a particular date], an active judge of the Court requested a poll on whether to rehear the case en banc. A poll having been conducted and there being no majority favoring en banc review, rehearing en banc is hereby DENIED .” The order itself, moreover, does not reveal the precise tally of the en banc poll. Thus, in cases where no dissenting opinions are filed, six active judges may still have voted in favor of en banc rehearing.

3.  Fed. R.App. P. 35(a).

4.  See id., Advisory Committee Notes (1998 Amendments).

5.  James L. Oakes, Personal Reflections on Learned Hand and the Second Circuit, 47 Stan. L.Rev. 387, 392 (1995) (emphasis supplied).

6.  Eisen v. Carlisle & Jacquelin, 479 F.2d 1005, 1021 (2d Cir.1973) (Kaufman, J., concurring in the denial of rehearing en banc).Judge Kaufman argued, in a related vein, that en banc proceedings threatened the “institutional integrity of the appellate court and the three-judge panel” because they send the message that “decisions reached by three-judge panels are not final, but represent merely one step on an elongated appellate ladder.” Irving R. Kaufman, Do the Costs of the En Banc Proceeding Outweigh Its Advantages?, 69 Judicature 7, 8 (1985); see also Green v. Santa Fe Indus., Inc., 533 F.2d 1309, 1310 (2d Cir.1976) (per curiam) (denying en banc review “not because we believe these cases are insignificant, but because they are of such extraordinary importance that we are confident the Supreme Court will accept these matters under its certiorari jurisdiction,” and stating that “[a] case in which Supreme Court resolution is inevitable should not be permitted to tarry in this Court for further intermediate action”).

7.  As recently as 2001, then-Chief Judge John M. Walker urged an approach to en banc review “that holds the process in reserve for the exceptional case that is an unlikely candidate for Supreme Court resolution.” John M. Walker, Jr., Foreword, 21 Quinnipiac L.Rev. 1, 14 (2001).

8.  Jon O. Newman, In Banc Practice in the Second Circuit: The Virtues of Restraint, 50 Brook. L.Rev. 365 (1984).

9.  By avoiding the costs and delays associated with convening en banc, Judge Newman argued that we can better “use judicial resources efficiently, concentrating our efforts on the prompt hearing and disposition of cases by panel opinion.” Jon O. Newman, In Banc Practice in the Second Circuit, 1989–93, 60 Brook. L.Rev. 491, 503 (1994). Other judges have likewise questioned the utility of en banc rehearings, insofar as they “produce[ ] either a majority opinion that was crafted in a purposefully vague manner to forge a consensus within the court, or a litany of diverging opinions, injecting a degree of uncertainty into the law.” Kaufman, supra note 6, at 8.

10.  Judge Newman observed twenty years ago that our limited approach to en banc rehearings has “contributed significantly to the high level of collegiality that this court enjoys.” Newman, supra note 9, at 503. At about the same time, Judge Oakes likewise reflected that “our en banc policy has helped us to maintain collegiality by avoiding the divisions that have caused friction on other courts of appeal.” Oakes, supra note 5, at 393. More recently, Judge Gerard E. Lynch stated that, although respect for a colleague's strongly held view may bring about acquiescence, it may be seen as an investment in collegiality. Gerard E. Lynch, Comment at the Columbia Law School Courts and Legal Process Workshop (April 23, 2012).

11.  Ricci v. DeStefano, 530 F.3d 88, 89 (2d Cir.2008) (Katzmann, J., concurring in the denial of rehearing en banc).

1.  The sleepiness was not attributable to the police but, rather, self-induced, purportedly by defendant's ingestion of Xanax pills shortly before arrest.

2.  A pre-trial services officer, who interviewed Taylor shortly after his second confession, reported him frequently falling asleep and needing to be roused. Nevertheless, the officer acknowledged that Taylor was able to provide the information necessary for the officer to complete his report to the court. See Tr. 319:19–320:11, 321:24–323:6, 325:16–22, S.A. 319–20, 321–23, 325. Meanwhile, a prison psychologist testified that prison records indicated that in admission interviews between his two confessions, Taylor was “vague” in responding to questions. Taylor II, 745 F.3d at 26 (quoting Tr. 110:14–24, S.A. 110). As that same witness testified, however, what Taylor was “vague” about was his mental health—a subject he may well have been reluctant to discuss with a prison psychologist. He was willing and able to provide coherent responses to questions seeking information about where he grew up, his family, education, and drug use. See Tr. 122:13–123:12, S.A. 122–23.

3.  Thus, while in Mincey v. Arizona, 437 U.S. 385 (1978), the Supreme Court concluded that hospitalization for serious physical injury, limited consciousness, incoherent responses, and unheeded requests to cease questioning compelled a legal conclusion of involuntariness, see id. at 396–402, after Mincey, we have repeatedly upheld voluntariness determinations by district judges who found that, despite being hospitalized, restrained, seriously injured, and medicated, persons were sufficiently lucid and coherent to make voluntary admissions to interrogating officers, see United States v. Siddiqui, 699 F.3d 690, 707 (2d Cir.2012) (upholding district court's voluntariness finding where, although defendant was hospitalized, restrained, in pain, and not administered Miranda warnings, she was “lucid and able to engage the agents in coherent conversation” and “agents' conduct was not overbearing or abusive”); United States v. Khalil, 214 F.3d at 121 (upholding district court's voluntariness finding where, although defendant had been shot, was in pain, and in hospital awaiting surgery, he was alert and responsive to agents' questions when making challenged statements); Campaneria v. Reid, 891 F.2d 1014, 1020 (2d Cir.1989) (upholding district court's voluntariness finding where, although defendant was in intensive care with knife wound, he was “alert and awake despite his pain,” and police had honored earlier requests to defer interview).

4.  As noted, Taylor was questioned for two to three hours, a circumstance that cannot be analogized to that in Blackburn.

5.  The proposed panel equation is unsupported by citation to any authority and flawed, in any event, by overstatement of the record. As already noted, the district court specifically credited testimony that Taylor was awake and lucid during his initial interrogation except for two or three occasions when he nodded off. Even then, verbal prompts were sufficient to allow him to regain focus. See supra at [3–4].

6.  Insofar as Taylor's interrogations took place at different sites—the first at FBI headquarters, the second at a courthouse—the panel appears to recognize that Elstad's venue factor affords no basis for identifying a continuing coercive effect. See Taylor II, 745 F.3d at 26.

7.  At the same time, Anderson recognizes that police trickery does not per se preclude a voluntary confession. See 929 F.2d at 99; see also Frazier v. Cupp, 394 U.S. 731, 739 (1969); Tankleff v. Senkowski, 135 F.3d at 244–45.

8.  We have repeatedly upheld the replacement of co-defendants' names with “neutral pronouns or references” as satisfactory to avoid Bruton error. United States v. Jass, 569 F.3d at 50; see United States v. Yousef, 327 F.3d 56, 149–50 (2d Cir.2003); United States v. Williams, 936 F.2d 698, 700–01 (2d Cir.1991); United States v. Tutino, 883 F.2d 1125, 1135 (2d Cir.1989). Many of our sister circuits have held similarly. See, e.g., United States v. Cone, 714 F.3d 197, 218 (4th Cir.2013); United States v. Green, 648 F.3d 569, 575–76 (7th Cir.2011); United States v. Vasilakos, 508 F.3d 401, 408 (6th Cir.2007); United States v. Taylor, 186 F.3d 1332,1336 (11th Cir.1999).

9.  The final redacted versions of Taylor's two confessions as admitted at trial are reproduced at the conclusion of this opinion as Appendices A and B.

10.  Although Tutino preceded the Supreme Court's decision in Gray v. Maryland, 523 U.S. 185, we have since confirmed that “our Tutino line of precedents remains useful” to Confrontation Clause analysis. United States v. Jass, 569 F.3d at 61.

11.  Miller was, in fact, caught on the pharmacy's surveillance video.

12.  Insofar as it appears that co-defendants were satisfied to have Miller identified by name in Taylor's confessions as part of a strategy bolstering their attack on her credibility, such a tactical decision would reach beyond forfeiture to demonstrate true waiver, negating all appellate review. See United States v. Quinones, 511 F.3d 289, 321 (2d Cir.2007).

Rosemary S. Pooler, Circuit Judge, concurs by opinion in the denial of rehearing en banc. José A. Cabranes, Circuit Judge, dissents by opinion from the denial of rehearing en banc. Reena Raggi, Circuit Judge, joined by José A. Cabranes, Richard C. Wesley, Peter W. Hall, Debra Ann Livingston, and Christopher F. Droney, Circuit Judges, dissents by opinion from the denial of rehearing en banc.